# Martechnext > Covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work. Public Ghost content for AI and LLM tooling. This file includes a bounded export of public pages first, then recent public posts. Append `.md` to any post or page URL to get the content in Markdown (for example, `/example-post.md`). ## Pages ### About MartechNext URL: https://www.martechnext.com/about/ Last updated: 2026-07-21T13:15:16.000Z ## What's next in marketing technology MartechNext covers the business of AI in marketing: who's building it, who's funding it, and how industries put it to work. Marketing technology is moving faster than ever. AI is rewriting how brands find, convert, and retain customers, and money is pouring into the companies making it happen. But most coverage stops at the press release. What does a funding round actually signal? Which tools genuinely work, and which are demos in a nice wrapper? And what does any of it mean for the marketer who has to deliver results next quarter? That's the gap MartechNext exists to fill. No hype, no rewritten press releases, but analysis grounded in real-world experience of building and implementing this technology. Here's what you'll find: - **Funding & Deals**: who's raising, who's buying, and why it matters - **AI in Marketing**: tools, use cases, and honest comparisons - **Travel**: a deep dive into one industry where martech meets a genuinely complex customer journey - **Analysis**: trends, acquisitions, and the occasional strong opinion - **Guides**: practical, hands-on material you can actually use Why travel as the first industry focus? Because it's the perfect stress test for marketing technology. Customer journeys that stretch over months, dozens of touchpoints, emotional decisions, and brutal competition on acquisition costs. If something works in travel, it will work almost anywhere. More industries will follow. ## Hi, I'm Bram ๐Ÿ‘‹ ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/Screenshot-2026-07-17-at-10.15.54.png) I've spent my career at the intersection of marketing, technology, and travel. Not as an observer, but hands-on: connecting fragmented data systems, building deep learning recommendation engines, running WhatsApp journey pilots, and helping travel and e-commerce brands make the shift from chasing volume to building customer value. Along the way I noticed something. The most interesting conversations I had were never about a specific tool. They were about the questions underneath: How do you translate complex data into human experiences? Where is the line between personalization and privacy? Why do most AI investments fail, and what do the successful ones do differently? Those are the questions I kept coming back to, and they're the reason MartechNext exists. I write about this space because I work in it. When I analyze a recommendation engine, a conversational commerce play, or a funding round, it's informed by having built and shipped this kind of technology myself, including the parts that didn't work. :) I don't have all the answers, and I'm suspicious of anyone who claims they do. The technology develops at breakneck speed, and what works today can be outdated tomorrow. That's exactly why I believe in sharing openly: the best ideas emerge from cross-pollination between different worlds. ## Every week: What's Next The heart of MartechNext is the weekly newsletter, **What's Next**. Every edition brings you: - One sharp analysis of the week - The latest funding rounds and deals in AI marketing - The martech news that actually matters, in short Straight to your inbox, free, personally written. No filler, no spam. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ## Let's talk Working on customer experience challenges in your own industry? Building something interesting in this space? Or simply want to exchange thoughts about the future of AI in marketing? I'd love to hear from you. You can reach me via [LinkedIn](https://www.linkedin.com/in/bramversteegh?ref=martechnext.com). ## Posts ### AI search skips your funnel: why 92% of AI traffic lands on decision-stage pages URL: https://www.martechnext.com/ai-search-skips-your-funnel-why-92-of-ai-traffic-lands-on-decision-stage-pages/ Last updated: 2026-07-28T09:52:23.000Z - Why WebFX's new study, analyzing nearly 600,000 AI-referred sessions, is worth your attention even before every methodological detail is public - What the 92% consideration/decision-stage figure actually implies about how people use ChatGPT, Perplexity, and Gemini to shop - Why this is a direct, real-world confirmation of the "conversation is the new funnel" thesis - The one caveat that genuinely matters (landing on a page is not the same as converting on it), and why it shouldn't stop you from acting - A practical to-do list to start capturing this traffic before your competitors do Picture a marketing director at a mid-sized B2B software company, staring at her analytics dashboard on a Tuesday morning. Organic search traffic to her blog is flat, has been for a year. But a new referral source keeps showing up in the source/medium report: chatgpt.com. Small numbers, growing fast, and oddly, these visitors aren't landing on the "10 tips for X" listicles that make up 80% of her content calendar. They're landing on the pricing page, the comparison page, the "X vs Y" post nobody on the content team was particularly proud of. If that pattern feels familiar, or if you've been quietly wondering the same thing while checking your own GA4, WebFX just gave it a name and a number. ## The claim: 92% of AI traffic skips the top of the funnel [WebFX analyzed nearly 600,000 AI-referred sessions](https://www.webfx.com/blog/ai/what-content-earns-ai-traffic/?ref=martechnext.com) and found that 92% of that traffic landed on consideration or decision-stage content, not on awareness-stage material like general blog posts or explainer content. Sit with that number for a second, because it's a genuinely interesting one. It suggests that when people ask an AI answer engine a question, the AI isn't sending them to your "what is X" starter content. It's routing them straight to the pages that help someone decide: comparisons, pricing, feature breakdowns, "best for" pages, product specifics. The stuff your content team probably ranks lower in the editorial calendar than the top-of-funnel awareness pieces, because awareness content is what "SEO best practice" has trained everyone to prioritize for the last fifteen years. If this pattern holds up as more data comes in, it's a real signal, not just a curiosity. It says AI referral traffic isn't a like-for-like replacement for organic search traffic. It behaves differently. It arrives further down the funnel, already having done its awareness-stage thinking somewhere inside the chat window, and it wants to compare, verify, and decide. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260728-1139-webfx-study-analyzes-nearly-600-000-ai-sessions-finds-92-of--visual-1.png) ****Where AI-referred traffic actually lands** โ€” 92% of nearly 600,000 AI-referred sessions landed on consideration or decision-stage pages, not awareness content. ยท Data: WebFX study, 2026 ## What WebFX actually measured, and what's still open I want to be straight with you about what we know and don't know here, because that matters for how much weight you put on this. WebFX is a digital marketing agency, and this study is also, functionally, content marketing for their own SEO and GEO services. That doesn't make the finding wrong, agencies sit on enormous amounts of real client data and are often first to spot these shifts precisely because they have visibility across hundreds of accounts. But it does mean I'd love to see the methodology laid out in more detail: which AI platforms were bucketed into "AI traffic" (ChatGPT only, or Perplexity and Gemini too), what time window the 600,000 sessions cover, and exactly how a page gets tagged as "awareness" versus "consideration" versus "decision" in the first place. [WebFX's own blog post is the primary source here](https://www.webfx.com/blog/ai/what-content-earns-ai-traffic/?ref=martechnext.com). I'd encourage you to read it directly rather than rely on secondhand summaries, including mine. Two other studies are circling the same question. Semrush has published research on ChatGPT clickstream behavior at scale, and Ahrefs has looked at which page types actually capture AI search traffic on its own site. Neither has been cross-checked against WebFX's funnel-stage breakdown in this piece, so I'm not going to pretend they corroborate the 92% figure. What I can tell you is that multiple independent teams are now pointing analytical firepower at the exact same question: what kind of content does AI traffic actually land on. That, by itself, tells you this isn't a fringe curiosity anymore. It's becoming a real measurement discipline. ## Why this confirms the thesis: the conversation is the new funnel Here's the part that connects to something I've been arguing on MartechNext for a while: the [conversation is the new funnel](https://www.martechnext.com/why-the-conversation-is-the-new-funnel-not-the-feed/). Commerce and research are moving into messaging threads and AI answer engines, and that shift changes where the real decision-making happens. The WebFX data, if it holds, is one of the clearest quantitative signals yet that this isn't theoretical. When someone types a question into ChatGPT, a meaningful chunk of the "top of funnel" thinking, the awareness stage, has already happened inside the conversation itself. The AI has effectively absorbed the "what is this category, what are my options" step. What it sends you, the actual referral, is someone who's already past that and wants to compare, verify, or buy. That's a genuinely useful thing to know, because it tells you where to put your effort. If your content strategy still treats top-of-funnel blog content as the front door and decision-stage pages as an afterthought buried three clicks deep, you may be optimizing for a customer journey that's quietly disappearing. The AI is becoming the front door. Your comparison page is becoming the landing page. This is also why feed quality and structured product data matter more than most marketing teams currently treat them. An AI engine can only cite your comparison page, your pricing table, or your feature matrix if it can actually parse it. That loops straight back to another MartechNext thesis: [the right order](https://www.martechnext.com/from-rising-acquisition-costs-and-data-chaos-to-ai-driven-growth-in-travel/). Structure the data first, then automate, then let AI make use of it. Sites with clean, well-tagged decision-stage content are the ones positioned to be cited when someone asks an AI "which of these is better for X." Sites with messy templates, PDF spec sheets, and pricing hidden behind a "contact us" form are not. ## The early-mover math Let's do a simple, round-numbers version of this, the kind of back-of-envelope math that tends to clarify a decision faster than a slide deck. Say a mid-sized SaaS company gets 20,000 monthly organic sessions today, split roughly evenly across awareness content (blog posts, guides) and decision content (comparison pages, pricing, case studies). AI referral traffic is currently a rounding error, maybe 300 sessions a month. If WebFX's 92% figure is even directionally right, roughly 276 of those 300 sessions are landing on decision-stage pages, a page type that historically converts at a meaningfully higher rate than a generic blog post. Now project that forward. AI referral traffic has reportedly been growing fast across multiple studies, even if the exact growth percentages differ by source and haven't all been independently verified. If that 300 becomes 3,000 within a year or two, and the funnel-stage skew holds, that's not "a bit more traffic." That's a new acquisition channel that arrives pre-qualified, landing exactly where you want prospects to land, at a moment when almost none of your competitors have bothered to optimize their decision-stage content for how AI engines actually read and cite it. That's the [early movers win channels thesis](https://www.martechnext.com/early-movers-into-chatgpt-ads-what-adwords-2001-can-teach-you-about-the-next-six-months/) playing out in close to real time. Every previous platform shift, AdWords in the early 2000s, Facebook ads in 2012, TikTok organic in 2019, WhatsApp Business messaging more recently, rewarded whoever built the foundation before the channel got crowded and expensive. GEO and AI-answer visibility looks like the same pattern, just earlier in the curve. I've made this argument before in the context of [travel brands disappearing from AI answers entirely](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/); this data gives it a sharper, more concrete edge: it's not just about being visible, it's about which pages you make visible. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260728-1139-webfx-study-analyzes-nearly-600-000-ai-sessions-finds-92-of--visual-2.png) ****The early-mover math on AI referral traffic** โ€” If the funnel skew holds as AI referral volume grows, a small trickle of sessions becomes a meaningful pre-qualified channel. ## The one caveat that actually matters I'm not going to pile on this study with a long list of nitpicks, because most of them are secondary to the real point. But there is one caveat worth holding onto, because it's the difference between acting smart and acting on vanity metrics. Landing on a decision-stage page is not the same as converting on it. "92% of sessions land on consideration/decision content" tells you about page-type distribution. It doesn't tell you about conversion rate, cost per acquired customer, or retention. [That's the volume-versus-value trap](https://www.martechnext.com/focus-on-value-not-volume-why-e-commerce-companies-should-care-more-about-their-ltv-cac-ratio/) I keep coming back to on this site : a per-session or per-channel statistic can look impressive while the total business impact stays modest, especially if the sample sizes involved (hundreds of AI-referred sessions, not tens of thousands, for most mid-sized sites right now) are still small enough that a handful of outlier conversions can swing the story. So treat "92% land on decision content" as a genuinely useful signal about where AI traffic behaves differently, not as proof that AI traffic is automatically more valuable traffic. The honest move is to track it separately in your own analytics and let your own conversion data settle the question, rather than importing someone else's aggregate number as your assumption. That's really the only caveat I'd put weight on here. Everything else, the exact methodology, the platform mix, whether Semrush and Ahrefs land on the same number, is worth watching but shouldn't stop you from acting today. Waiting for a perfectly footnoted study before you touch your decision-stage content is how you end up optimizing for a funnel that no longer exists. ## What to do about it: a marketer's checklist Here's the practical part. If this pattern holds even partially true for your traffic, here's where I'd start: 1\. **Segment AI referral traffic in your analytics now.** Set up a dedicated channel grouping for chatgpt.com, perplexity.ai, gemini.google.com, and copilot.microsoft.com referrals in GA4 or your analytics platform of choice. You can't act on a pattern you can't see. 2\. **Audit your decision-stage content inventory.** List every comparison page, pricing page, "alternatives to X" post, and feature matrix you have. Most teams are surprised by how thin this list is compared to their awareness-stage blog archive. 3\. **Fix the pages AI engines can't parse.** Pricing hidden behind gated forms, specs buried in PDFs, comparison tables rendered as images: none of that is citable by an AI answer engine. Structured, crawlable, plain-text decision content wins. 4\. **Write for extraction, not just for humans skimming.** Direct answers to direct comparison questions ("X vs Y: which is better for Z") get cited more easily than narrative prose. Add clear headers, short direct answers up top, and structured data where relevant. 5\. **Shift a slice of your content budget from awareness to decision.** Not all of it. But if 92% of your AI-referred visitors are already past the awareness stage, pouring more resources into another "beginner's guide to X" is diminishing returns. 6\. **Measure conversion by referral source, not just sessions.** This is where you test the caveat above. Does your AI-referred traffic actually convert better, worse, or the same as organic and paid? Don't assume, check. 7\. **Start tracking your brand's citations inside AI answers**, using whatever GEO monitoring tooling is available to you. Being present in the conversation is the leading indicator; traffic is the lagging one. None of these require a large budget or a new hire. They require someone on your team deciding this is worth an afternoon this month, not a "maybe next quarter" item. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260728-1139-webfx-study-analyzes-nearly-600-000-ai-sessions-finds-92-of--visual-3.png) ****A marketer's checklist for AI-referred traffic** โ€” Seven concrete steps to start capturing decision-stage AI traffic this month. ## Verdict I like this study more than I distrust it. The 92% figure comes with real methodological gaps that deserve to be closed before anyone treats it as gospel, and I'd rather see WebFX's full sample details than take the headline number on faith. But the underlying pattern it points at, AI referral traffic skewing hard toward decision-stage content, fits everything else I'm seeing about how people actually use ChatGPT and Perplexity to shop, and it fits a thesis I've been building on for a while: the conversation is genuinely becoming a funnel stage of its own, not just a curiosity metric in your analytics. My take: don't wait for a peer-reviewed version of this study to show up. Audit your decision-stage content this month, make sure AI engines can actually read it, and measure your own AI-referred conversion data honestly. If it turns out your AI traffic converts as well as WebFX's data implies, you've gotten ahead of a channel while it's still cheap to win. If it turns out your numbers look different, you've lost an afternoon, not a quarter. That's a good trade either way :) Wrestling with GEO or AI-referral measurement in your own stack? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Marketing Before the Search Bar: What Predictive Intent Means for the Conversation Economy URL: https://www.martechnext.com/marketing-before-the-search-bar-what-predictive-intent-means-for-the-conversation-economy/ Last updated: 2026-07-28T07:34:38.000Z **In this article:** - Why "predictive intent" matters if the conversation, not the search bar, is becoming the funnel - What's actually new here versus what's a rebrand of existing ABM and intent-data tooling - The data precondition nobody skips past, no matter how good the model is - The Target pregnancy story, and why it's the cautionary tale this category needs - A practical, non-hyped roadmap for testing predictive intent without creeping out your customers - My verdict: cautiously interested, evidence still pending --- Picture a mid-sized outdoor retailer's marketing manager looking at a new dashboard row: a household flagged as "78% likely to buy ski gear in the next four weeks," generated before anyone in that household has typed a single search query, opened an email, or clicked an ad. No keyword to bid on. No search intent to capture. Just a model saying: this person is about to want something, reach them now. That's the pitch behind "predictive intent," a phrase getting airtime in martech trade coverage this year, including a recent piece from Martech Series arguing marketing is shifting from reactive, search-triggered targeting to models that anticipate need before it surfaces as a query. It's a good story. The question worth asking, as always, is whether the technology backs it up, and what happens to the rest of our search-visibility playbook if it does. ## Why this matters right now I've spent a fair amount of space on this site arguing that the conversation is becoming the new funnel: commerce moving into WhatsApp threads, Meta Business Agents, and ChatGPT product answers instead of ten blue links (see the [Meta Conversations 2026 ](https://www.martechnext.com/meta-conversations-2026-the-business-agent-is-here-and-it-changes-your-whatsapp-strategy/)piece. [The GEO playbook](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/) I've written about is essentially SEO's successor: getting your brand surfaced correctly when someone asks an AI assistant a travel or shopping question. But all of that still assumes a query happens. Someone types something, or says something, and then the system responds. Predictive intent proposes skipping that step entirely: model the need before the question exists, and act on it. If that's technically real, and reliably so, it's not a footnote to the GEO thesis. It's a layer that sits in front of it. Your product feed still needs to be clean for [ChatGPT to recommend you](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/). Your WhatsApp journey still needs to [convert once someone's in the thread](https://www.martechnext.com/whatsapp-marketing-in-2026-what-marketers-need-to-know-to-use-it-effectively/). But before either of those matters, a predictive layer decides whether you get a shot at the conversation at all. That's a genuinely exciting proposition for marketers tired of bidding on the same high-intent keywords everyone else is bidding on. Reaching someone before the auction even opens is, in theory, the cheapest possible acquisition channel there is. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260728-0917-predictive-intent-and-martech-marketing-to-customers-before--visual-1.png) ****The funnel is moving earlier** โ€” Predictive intent proposes acting before a query exists, ahead of conversation, GEO, and conversion ## What's actually new here, and what isn't Here's where I want to slow down, because this is exactly the kind of claim that deserves scrutiny before adoption. Intent data as a category isn't new. Vendors like 6sense, Demandbase, Bombora, and TechTarget's Priority Engine have been selling account-level "surge" and propensity scores for years, built by aggregating first- and third-party behavioral signals into a model that flags which accounts are showing buying signals before a rep or a marketer notices (per Demandbase and DemandScience's own explanations of how their intent data works). That's the mechanic ABM has run on for a decade. So what, specifically, does "predictive intent" add on top of that? Based on what I could verify, the honest answer is: it's unclear. The Martech Series piece and adjacent trade coverage describe the concept in aspirational terms, marketing before the search happens, but I found no source that draws a clean technical line between "predictive intent" and the propensity scoring that's already commercially available. It's entirely possible this is a genuinely new modeling approach. It's also entirely possible it's a rebrand riding the AI wave, the same way "smart" got attached to every product category around 2015. I'll say the quiet part: I couldn't verify any independent, vendor-neutral figures on accuracy, conversion lift, or ROI specific to "predictive intent" as its own category. If a vendor pitches you a number here, ask what it was measured against, what the sample size was, and whether anyone outside the vendor validated it. Vendor-run benchmarks are directionally interesting. They are not proof. That's not a reason to dismiss the idea. It's a reason to treat the label skeptically while taking the underlying capability seriously. ## The data precondition nobody gets to skip Assume, for a moment, that predictive intent models genuinely work better than existing propensity scoring. They'd still need the same thing every AI system needs to function: clean, unified, structured behavioral data flowing from every touchpoint into one place. This is the thesis I keep coming back to on this site: [**the right order is data, then automation, then AI**](https://www.martechnext.com/from-rising-acquisition-costs-and-data-chaos-to-ai-driven-growth-in-travel/). MIT's widely cited figure that roughly 95% of enterprise AI pilots fail isn't because the models are bad. It's because most organizations are trying to run sophisticated prediction on top of what I've called data spaghetti, disconnected systems, inconsistent identifiers, half-synced CRMs. Predicting intent before a search happens is, structurally, a harder problem than scoring an account that's already showing surge activity. It needs more signal, cleaner joins across channels, and a feedback loop that tells the model when it's wrong. If a vendor is pitching predictive intent without first talking about the data foundation underneath it, that's worth noticing. There's also an open question I want to flag honestly rather than paper over: how third-party cookie deprecation and GDPR, CCPA, and CPRA constraints actually affect the data inputs this kind of model needs. I looked for a solid, current answer on this and didn't find one I'd stand behind. Any vendor claiming their predictive model is "privacy-safe by design" deserves a direct follow-up question: safe under which regulation, using which consent basis, for which data types. Don't accept the phrase at face value. ## The creepy line, and why it matters more than the accuracy number The most famous story in this space is Target's pregnancy-prediction model, the one where a father allegedly received maternity coupons addressed to his teenage daughter before she'd told her family she was pregnant. It's the go-to anecdote every predictive marketing article reaches for, including, I'd guess, some future draft of this exact piece if I hadn't stopped myself. I want to use it carefully. The exact details of that story have been disputed and likely embellished across a decade of retellings (multiple accounts diverge on specifics). Treat it as illustrative of a risk, not as a clean, verified case study. But the underlying dynamic is real regardless of the exact anecdote: a model that's accurate enough to predict something a customer hasn't disclosed yet can produce a genuine trust rupture, even when it's technically working as designed. This is exactly why **automation needs human accountability layered on top of it**. The model can flag the propensity. A person, or at minimum a rules layer with real judgment built in, should decide how and whether that gets acted on in a customer-facing way. Say a travel brand's predictive model flags 10,000 households as high-propensity for a ski trip, at roughly 60% real accuracy. That means around 6,000 of those people were probably going to book anyway, so the incremental value is smaller than the headline number suggests, and the other 4,000 get an eerily specific offer for a trip they had no plans to take. Best case, that's a mildly odd email. Worst case, it's the moment a customer decides your brand is watching too closely. The lift you can measure is never the whole story. The trust you spend to get it usually isn't on the same spreadsheet. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260728-0917-predictive-intent-and-martech-marketing-to-customers-before--visual-3.png) ****The gap between headline accuracy and real lift** โ€” At 60% accuracy on 10,000 flagged households, most would have booked anyway, and thousands get an offer they never asked for. (illustrative example) ## Who's actually positioned to win this If predictive intent turns out to be a real, distinct capability rather than a rebrand, this is a textbook case of **early movers winning the channel**, the same pattern I've tracked across AdWords, Facebook, TikTok, and WhatsApp. Being early to a genuinely new targeting layer, before it gets commoditized and priced up, is usually worth more than waiting for the category to mature. But "early mover" only pays off if the underlying capability is real. Being early to a rebrand just means you paid full price for something you already had. I don't have a confirmed list of vendors shipping a product specifically labeled "predictive intent" with disclosed methodology (source needed, and I'd genuinely want to see one before recommending a specific tool). Until that exists, the honest move is to build the muscle in-house on data you already control and have consent for, rather than buying a black-box score from a vendor who can't explain how it's built. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260728-0917-predictive-intent-and-martech-marketing-to-customers-before--visual-2.png) ****Predictive intent: verified vs. vendor claim** โ€” What's established in intent-data tooling versus what remains an unverified vendor claim about predictive intent specifically. ## A practical way to test this without the hype You don't need to wait for the category to settle to start learning something useful. Here's how I'd approach it: 1\. **Audit your first-party data before anything else.** If your CRM, booking engine, and email platform don't share a clean customer ID, you're not ready for predictive scoring, you're ready for step one of thesis 2. 2\. **Start with propensity scoring you already have consent for.** Existing ABM or CRM-native scoring (Salesforce, HubSpot, and similar platforms already offer this) is a legitimate, lower-risk place to test the underlying idea before buying a new "predictive intent" label. 3\. **Run it as an internal signal first, not a customer-facing trigger.** Let your sales or lifecycle team see the flag before any automated message goes out. That's the human accountability layer calls for. 4\. **Wire any validated signal into the channels where the conversation actually happens**, your WhatsApp journeys, your chat interfaces, [not just legacy email blasts](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/). 5\. **Holdout test everything.** Split your flagged high-propensity group, treat half, leave half untouched, and measure the actual incremental lift, not the flattering per-recipient number. This is the same discipline I've argued for in the [LTV/CAC](https://www.martechnext.com/focus-on-value-not-volume-why-e-commerce-companies-should-care-more-about-their-ltv-cac-ratio/) piece: totals and incrementality, not vanity multipliers. If you can't get a vendor to walk you through their methodology at the same level of detail you'd expect from your own holdout test, that's your answer for now. ## My verdict Predictive intent is a genuinely interesting direction, and it fits cleanly into where I think the funnel is heading: earlier, more individual, and less dependent on a visible query. If it works, it's the natural front layer to everything I've written about GEO and conversational commerce. But right now, it's a compelling concept sitting ahead of its evidence. I couldn't find independent proof that it's technically distinct from the propensity scoring ABM vendors have sold for years, and I couldn't find validated accuracy or ROI figures that weren't vendor marketing copy. That doesn't mean it's fake. It means it's unproven, which is a different thing entirely. My advice: build the data foundation now, since you'll need it either way. Pilot the concept internally with data you already have consent for. And treat any vendor pitching "predictive intent" as a distinct, priced-up category the way you'd treat any bold new label: ask for the methodology before you ask for the demo. Are you seeing "predictive intent" show up in vendor pitches on your desk already? I'd love to exchange notes on what they're actually claiming versus what they can show. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Can you actually buy a flight by talking to Alexa? Amazon is about to find out URL: https://www.martechnext.com/can-you-actually-buy-a-flight-by-talking-to-alexa-amazon-is-about-to-find-out/ Last updated: 2026-07-28T07:02:56.000Z **In this article:** - Why Amazon picked travel, of all categories, to test in-chat purchasing - What's actually been confirmed versus what's still press language - Why travel is the hardest possible test case for conversational commerce - The reliability problem nobody wants to talk about yet - Who actually wins if this works: Amazon, Priceline, or nobody - A practical read on what travel marketers should do right now, not in six months Picture someone lying on the couch on a Tuesday night, telling Alexa they need a flight to Lisbon for a long weekend in October. No app, no ten browser tabs, no fifteen-minute price comparison ritual. Alexa checks dates, pulls a price, and asks: shall I book it on your saved card? That moment, the one where a voice assistant closes a real purchase instead of just answering a question, is the moment every conversational AI platform has been chasing and none of them have convincingly nailed. Amazon just said, in effect: we're going to be the ones who crack it, and we're starting with travel. According to Skift, Amazon's newly announced Model Context Protocol (MCP) support for Alexa+ has pulled in Priceline, Viator, Virgin Atlantic, Lyft, Brightline, FlixBus, and Greyhound as early partners building booking tools for the assistant. A new "Amazon Wallet" feature would let people pay using stored cards, right there in the conversation. Skift frames it plainly: Amazon is betting that stored payment credentials, MCP, and 180 million Prime members can solve the problem that stopped ChatGPT cold, namely getting people to actually complete a purchase inside an AI chat. That's a big claim. It's also, if it works even partially, a live test of one of the ideas I keep coming back to on this site: **the conversation is the new funnel**. This is that thesis with its sleeves rolled up, tested against one of the hardest, highest-stakes purchase categories there is. ## What's actually been shown, and what's still a press release Let's separate signal from noise, because there's a lot of noise here. Confirmed, per Skift's reporting: Amazon has MCP support live for Alexa+, a list of named launch partners including Priceline, and a wallet feature for stored-card payments. That's real. Amazon is also not doing this in isolation. Earlier reporting from Skift (April 2026) already flagged Alexa+'s travel planning and booking ambitions, and PPC Land reported that Alexa+ has separately added Expedia, Yelp, Angi, and Square integrations for travel and home tasks. AltexSoft has also written about Amazon bringing travel booking to Alexa using Expedia's inventory. Not yet confirmed, at least not in anything I've been able to verify: the actual end-to-end mechanics. Does a Priceline booking inside Alexa+ complete the full search-price-confirmation flow natively, or does Alexa hand you off to Priceline's app or site at some point in the process? Skift's language ("trying to sell trips inside a chat") is doing a lot of work here, and it's the kind of phrasing that can describe anything from a fully agentic checkout to a glorified deep link with a friendlier voice. I'd also flag that the commercial terms between Amazon and Priceline (referral fees, revenue share, exclusivity) haven't been disclosed anywhere I can find *(source needed)*. And whether the separate Expedia integration is a competing bet or a parallel, hedged track is genuinely unclear from the reporting available. Amazon may simply be playing the field, letting multiple OTAs build on the same rails and seeing who converts. None of that is a knock on the ambition. It's just the difference between "Amazon announced a capability" and "Amazon proved people will book flights this way." Those are two very different stories, and right now we only have the first one. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260728-0854-amazon-s-alexa-is-trying-to-sell-trips-inside-a-chat-priceli-visual-1.png) ****The conversational booking stack: confirmed vs. still a claim** โ€” What Amazon's Alexa+ travel push has actually confirmed, versus what remains press language. ยท Data: Skift, PPC Land, AltexSoft, 2026 ## Why travel is the hardest test case Amazon could have picked If Amazon wanted an easy win to prove conversational commerce works, travel is close to the last category I'd have chosen. And that's exactly why it's the right one to watch. Think about what a flight or hotel booking actually demands compared to, say, reordering paper towels. Dates have to be exact. Cancellation policies matter. Prices move by the hour. A wrong booking isn't a minor inconvenience, it's a ruined trip or a fight with customer support. This is the category with the least tolerance for the kind of "roughly right" answer an AI assistant is often happy to give you. It's also, not coincidentally, one of the categories with the worst acquisition economics in all of commerce. Travel's customer acquisition costs have climbed roughly 35% over the past three years while lifetime value has grown only about 4.5% in the same period. That's thesis #1 territory: **value over volume**, and it's exactly why a new, potentially cheaper booking channel is so tempting for travel brands right now. If Alexa+ can shave even a sliver off the cost of acquiring or retaining a traveler, [every OTA and airline on that early partner list has a reason to lean in hard](https://www.martechnext.com/from-rising-acquisition-costs-and-data-chaos-to-ai-driven-growth-in-travel/). But here's the tension. High-CAC categories are exactly where mistakes are most expensive, and where the trust bar for a new channel is highest. Nobody's cutting corners on a โ‚ฌ40 grocery reorder. Plenty of people will absolutely bail on a โ‚ฌ600 flight booking the moment the assistant hesitates, mishears a date, or can't clearly explain the cancellation terms. That tension, high upside if it works, high cost if it doesn't, is the whole story here. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260728-0854-amazon-s-alexa-is-trying-to-sell-trips-inside-a-chat-priceli-visual-2.png) ****Travel's acquisition squeeze: CAC up, LTV flat** โ€” Rising acquisition costs and flat lifetime value make travel brands eager for cheaper booking channels. ## The reliability problem nobody's advertising Amazon's framing, understandably, is confident: stored credentials, MCP, 180 million Prime members, problem solved. But the on-the-ground reporting on Alexa+ itself tells a slightly different story. Independent beta testers reportedly described the pre-launch Alexa+ experience as "unbearably erratic," per Men's Journal. Consumer Reports' hands-on review of Alexa+ was positive on the AI assistant itself but flagged the surrounding app experience as a drag on usability. Neither of those reports is specifically about travel booking, and I want to be careful not to overstate that connection. But they're a useful counterweight to the launch-partner press release, and they matter for a simple reason: general-purpose reliability problems tend to show up hardest in the highest-stakes tasks, not the lowest-stakes ones. This is thesis #5 in action: **automation with human accountability**. Booking a flight is exactly the kind of task where you want the AI to do the tedious part (searching, comparing, holding the details in its head) and a human, whether that's the traveler themselves or a confirmation step, to do the final judgment call. The trust and control curve lags the capability curve, and nowhere is that gap more visible than in a system that's willing to spend your money on your behalf. Let's be honest: an assistant that's "unbearably erratic" reordering a book is annoying. The same erraticism applied to a multi-leg flight booked on stored payment credentials is a different category of problem entirely. ## Who actually wins if this works Assume, for a moment, that Amazon gets the mechanics right and people genuinely start booking travel through Alexa+. Who benefits? The obvious answer is Amazon, and it's not wrong. Owning the conversational interface means owning the first and last touchpoint of the booking journey, which is exactly the kind of position that generates the data flywheel described in thesis #4\. Every booking, every hesitation, every "actually, book the earlier one instead" teaches the assistant something, and that learning compounds. Whoever sits at that layer accrues the relationship data, and right now that's clearly Amazon, not Priceline. Priceline's upside is more modest but still real: distribution into 180 million Prime households without having to build the conversational layer themselves. That's the same logic that made early advertisers pile into AdWords, or early brands pile into WhatsApp Business, back when those channels were unproven but the audience access was undeniable. Thesis #7, **early movers win channels**, applies directly here. If Priceline, Viator, and the transport players on that list get the booking flow and feed formatting right before their competitors even have an MCP server running, they could lock in a real head start once (or if) volume shows up. What's conspicuously absent from that early partner list is instructive too: the major hotel chains, the big US airlines, Booking.com, Expedia's core brand. That's either a sign those players are waiting to see if this works before committing engineering resources, or a sign they're wary of handing Amazon the transaction relationship and the data that comes with it. I'd guess it's mostly the former with a healthy dose of the latter. Nobody wants to rebuild their booking flow around a channel that might not survive contact with real users. ## What travel marketers should actually do about this You don't need to have an MCP server running by Friday. But you also shouldn't file this under "wait and see" and forget about it, because thesis #7 exists precisely because the brands that waited on AdWords, on Facebook, on WhatsApp, all paid more to catch up later. Here's a practical, unglamorous roadmap: 1\. **Audit your booking flow for machine legibility.** Can your availability, pricing, and cancellation policy be read cleanly by an external agent today? If your rates only live inside a JavaScript-heavy booking widget, no assistant is going to surface them accurately, MCP or not. 2\. **Get your product feed in order before you get your Alexa strategy in order.** This is the same [discipline GEO and AI](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/) shopping feeds demand elsewhere. Structured, accurate, current data is the entry ticket, not a nice-to-have. 3\. **Decide, deliberately, whether you want to be found via Amazon at all.** There's a real strategic question about whether you want your transaction data and customer relationship intermediated by Amazon, versus building your own conversational channel where you own that relationship end to end. Full disclosure, as always: I'm the founder of Zoey, a WhatsApp journey automation platform for the travel industry, so I have skin in this exact game, owning the conversational relationship directly rather than renting it from a platform. 4\. **Watch conversion, not headlines.** The story worth tracking isn't "Amazon launches feature." It's whether actual booking volume and actual CAC move, six to twelve months from now. Nobody has published that data yet, and until they do, everything here is directional. 5\. **Don't assume voice and chat convert the same way.** A screen lets people compare, scroll, second-guess. A voice conversation compresses that into a much faster yes/no. That could be an advantage (less friction) or a liability (less confidence), and right now nobody outside Amazon knows which. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260728-0854-amazon-s-alexa-is-trying-to-sell-trips-inside-a-chat-priceli-visual-3.png) ****Trust vs. capability: where conversational commerce sits by task** โ€” High-stakes, high-ambiguity bookings sit furthest from where AI assistants can be trusted to close the loop today ## The verdict I don't think Amazon has solved conversational commerce for travel yet, and I don't think the reporting so far lets anyone credibly claim they have. What Amazon has done is assemble the right ingredients: distribution (Prime), payment infrastructure (Wallet), a technical standard (MCP), and, crucially, partners willing to be the guinea pigs. That's a meaningfully stronger starting position than OpenAI had when its own in-chat shopping ambitions stalled. But travel is a brutal proving ground precisely because it punishes the gap between capability and reliability harder than almost any other category. "Unbearably erratic" and "here's your confirmed flight to Lisbon" don't coexist for long before someone loses money or misses a connection. My honest read: this will work, eventually, for a narrower slice of travel than the press coverage implies. Simple, low-ambiguity bookings (a known route, a known hotel, a repeat traveler with clear preferences) are where I'd expect real conversion first. Complex, multi-leg, first-time trips are where I'd expect the assistant to still hand off to a human or a screen for a long while yet. If you're a travel marketer, the mistake isn't ignoring this. It's either overreacting (rebuilding your stack around Alexa+ before there's conversion data to justify it) or underreacting (assuming voice commerce is a novelty and skipping the groundwork). Get your data and feeds in shape now. Decide who you want to own the relationship with your traveler. Then watch, for real, whether people actually say yes when Alexa asks for the card. Wrestling with where conversational channels fit into your travel marketing stack? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Early movers into ChatGPT Ads: what AdWords 2001 can teach you about the next six months URL: https://www.martechnext.com/early-movers-into-chatgpt-ads-what-adwords-2001-can-teach-you-about-the-next-six-months/ Last updated: 2026-07-27T12:12:56.000Z **In this article:** - What OpenAI actually shipped this week, and which parts are genuinely new versus repackaged - Why the speed of ChatGPT Ads' maturation curve matters more than any single feature - The tension between new measurement tools and OpenAI's own admitted measurement gaps - How the $100M ARR claim stacks against independent forecasts, and what that gap tells you - A practical way to think about testing budget here without betting the farm - Why this connects to the oldest lesson in performance marketing: early movers win channels Picture a mid-sized DTC brand's paid media lead, scrolling Search Engine Land on a Friday afternoon, seeing the headline "ChatGPT Ads adds conversion bidding, geo exclusions and bulk campaign tools" and thinking: another platform update, file it under "check later." That instinct is understandable. It's also, I think, wrong. Not because the individual features are groundbreaking. Because of how fast they arrived. ## What actually shipped Let's get the facts straight first, because there's a lot of noise around this one. According to Search Engine Land, OpenAI's update to ChatGPT Ads includes conversion-optimized campaigns (oCPC), average daily budgets on a rolling seven-day model, automatic budget pacing, geographic exclusions, AppsFlyer and Adjust mobile measurement integrations, Automatic Advanced Matching (hashed PII matching, the same concept Meta and Google use for match rates), and an asynchronous Bulk API for managing campaigns, ad groups and ads at scale. None of this is exotic if you've run Google or Meta campaigns. Geo exclusions, daily budget pacing, bulk campaign management: this is table stakes on mature platforms. What's notable is that ChatGPT Ads didn't have any of it a few months ago. Some of these are genuinely new capabilities (oCPC bidding, the Advanced Matching layer). Others are OpenAI catching up to features Google Ads shipped over a decade ago (bulk editing, geo exclusions). Worth being honest about that split: this is less "OpenAI invents advertising" and more "OpenAI builds the boring plumbing every mature platform needs." ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260727-1339-chatgpt-ads-adds-conversion-bidding-geo-exclusions-and-bulk--visual-1.png) ****ChatGPT Ads: five months from pilot to platform** โ€” OpenAI compressed years of typical ad-platform maturation into five months. ยท Data: Search Engine Land, 2026 ## The real story is the speed, not the features Here's the part that made me sit up. In February 2026, ChatGPT Ads launched as an invite-only pilot with a minimum spend somewhere around $200,000 to $250,000\. By April, that minimum had dropped to $50,000\. On May 5, 2026, OpenAI opened self-serve Ads Manager to all US businesses with a $0 minimum. Nine days later, custom audience upload arrived for suppression and retargeting. By late May, CPA-style conversion bidding started rolling out to select advertisers, reaching all accounts with pixel or Conversions API setup by early June. And now, in July, we get bulk tools and Advanced Matching. That's roughly five months from "enterprise pilot with a six-figure floor" to "self-serve platform with conversion bidding, mobile measurement partner integrations, and bulk campaign management." Compare that to how long it took Google Ads (then AdWords) to get from launch to conversion tracking and automated bidding. Or how long Meta took to go from boosted posts to Advantage+ automation. Those maturation curves ran in years, not months. This one is compressing hard, and that compression is the actual news here. This is thesis #7 in action: **early movers win channels**. Every platform shift, AdWords in the early 2000s, Facebook ads in 2007, TikTok ads in 2019, WhatsApp Business messaging more recently, rewarded the advertisers who built the foundation before the channel matured and CPCs caught up to competitive reality. Right now, OpenAI is recommending a starting max bid of $3 to $5 per click on a relevance-weighted second-price auction. Early pilot CPMs cleared as low as $25, against a $60 default max, and one leaked agency deck reportedly offered CPMs as low as $15\. Those numbers will not last. They never do. AdWords CPCs in 2001 were pennies. They aren't pennies now. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260727-1339-chatgpt-ads-adds-conversion-bidding-geo-exclusions-and-bulk--visual-3.png) ****The CPM window: early pilot pricing vs. today's default** โ€” Early ChatGPT Ads CPMs sat well below the current default ceiling, a gap that historically closes fast The arbitrage window on any new ad channel is a function of how few advertisers understand the platform relative to how many users are already there. ChatGPT reportedly has 800 million users. Most of them have never seen an ad in the product, because ads currently only reach Free and Go tier users in the US, Canada, Australia and New Zealand, with the UK, Japan, South Korea, Brazil and Mexico coming next. That's a big audience and a comparatively tiny number of advertisers bidding for attention inside it, at least for now. **The window doesn't open twice. It just closes.** [Meta Business Agents: buying in the chat](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/) [Meta Conversations 2026: the Business Agent is here](https://www.martechnext.com/meta-conversations-2026-the-business-agent-is-here-and-it-changes-your-whatsapp-strategy/) ## The measurement tension nobody's resolved yet Now for the part that deserves proportionate skepticism, because it genuinely matters. Search Engine Land has separately reported that OpenAI's ad platform has struggled to show advertisers clear proof that their spend is working. That's not a minor footnote. It sits in direct tension with this week's update, which is explicitly marketed as a measurement upgrade: Advanced Matching, MMP integrations, conversion bidding. You can't optimize toward conversions you can't reliably measure, and you can't build advertiser trust with bidding sophistication if the underlying attribution is still shaky. This maps straight onto thesis #2, **the right order**: structure your data first, then automate the repetitive, only then apply AI. Advanced Matching and pixel/CAPI integrations are OpenAI trying to fix the data-plumbing gap retroactively, after already shipping AI-driven bidding on top of it. That's backwards from the recipe, and it's worth remembering that MIT research puts the failure rate of enterprise AI initiatives at 95%, largely because they're built on messy data foundations rather than clean ones. OpenAI isn't exempt from that risk just because it built the model. None of the specifics here (conversion lift, Advanced Matching match rates, MMP accuracy) have been independently validated. Every number traces back to OpenAI's own product descriptions, relayed through trade press. That's not an accusation of dishonesty, it's just how early-platform reporting always looks: directionally credible, not gospel. Treat it the way you'd treat any vendor-run benchmark. ## The $100M question OpenAI's ad program reportedly hit $100 million in annualized revenue within six weeks of launch, with over 600 companies enrolled, a figure a company representative confirmed and Business Insider corroborated independently. That's a genuinely fast ramp for a brand-new ad product. But EMARKETER expects total US chatbot advertising revenue to stay under $1 billion for all of 2026, and one analyst has pointed out that OpenAI needs substantially more user adoption and ad volume to hit its own internal targets. Both things can be true. $100M in annualized run-rate after six weeks is impressive momentum. It's also a rounding error next to Google's or Meta's ad businesses, and a sub-$1B total category forecast for the year tells you this channel is still small in absolute terms, even if it's growing fast in relative terms. Don't let "$100M ARR" make you think ChatGPT Ads is already a mainstream channel. It isn't. It's an early one. That's exactly the point. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260727-1339-chatgpt-ads-adds-conversion-bidding-geo-exclusions-and-bulk--visual-2.png) ****ChatGPT Ads: hype metric vs. category reality** โ€” A fast six-week ramp still sits inside a small overall category forecast. ยท Data: Business Insider, EMARKETER, 2026 ## What advertisers should actually do with this I'd steer clear of two overreactions here. One is "this is nothing, check back in 2027." The other is "drop 30% of your Google budget into ChatGPT Ads tomorrow." Neither is the right move. Here's a more useful way to think about it, tied to thesis #1, **value over volume**: 1\. **Treat this as a test-and-learn channel, not a scale channel, for now.** Run a small, deliberately-sized pilot (think holdout-test discipline, not "spend until it stops working"). You're paying for optionality and channel knowledge, not immediate ROAS. 2\. **Get your pixel and Conversions API instrumentation in order before you touch conversion bidding.** oCPC only works as well as your conversion signal. If your data plumbing is spaghetti, don't expect an AI bidding layer to fix it, that's the exact trap thesis #2 warns about. 3\. **Measure by LTV:CAC and incrementality, not raw conversion count.** With CPA-style bidding now live, it's tempting to chase the volume metric the dashboard shows you. Resist it. The same discipline that applies to Google and Meta applies here: a cheap conversion that [churns in a week isn't a win](https://www.martechnext.com/focus-on-value-not-volume-why-e-commerce-companies-should-care-more-about-their-ltv-cac-ratio/). 4\. **Remember oCPC is optimized cost-per-click, not true pay-per-conversion.** OpenAI is still charging you per click while predicting conversion likelihood behind the scenes. Don't let the naming imply you only pay when someone converts. 5\. **Build your feed and off-site measurement discipline now, while the audience-to-advertiser ratio is still generous.** This is the same principle behind [GEO for organic visibility](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/): the brands that show up cleanly and get measured accurately before the channel matures are the ones who compound that advantage once CPCs rise and the auction gets crowded. ## How this stacks against the incumbents I want to be careful here, because the honest answer is: we don't yet have solid comparative performance data. What we do know is structural. ChatGPT Ads targets contextually, based on what's actually being discussed in the conversation, not demographically. There are no match types, no Quality Score, no audience segments in the traditional PPC sense, though custom audience upload for suppression and retargeting arrived shortly after the May self-serve launch. That's a fundamentally different targeting model than Google's Performance Max or Meta's Advantage+, both of which lean hard on algorithmic audience expansion across demographic and behavioral signals. Whether contextual-only targeting scales as well as those systems is genuinely an open question. Nobody has published a clean comparison yet, and I'd be skeptical of anyone claiming otherwise this early. What I can say is this: the whole thing is evolving in the direction of a full-funnel commerce layer, not just an ad slot. OpenAI is testing multi-advertiser placements and a shopping product carousel with Etsy and Shopify checkout integration. That's the conversation becoming the funnel, quite literally: discovery, ad, and checkout in the same thread. This is thesis #6 made concrete. **The thread is the new funnel, and OpenAI is racing to build the toll booth before the traffic arrives.** ## Open questions worth watching A few things I'd keep an eye on over the next two quarters: - Whether independent agencies publish real conversion-lift and Advanced Matching match-rate data, rather than OpenAI's own descriptions - Whether the exact rollout sequence for conversion bidding gets a clean, primary-source timeline from OpenAI (reporting has been inconsistent across outlets) - Whether the multi-advertiser placement test and shopping carousel actually convert, given OpenAI's earlier, publicly walked-back struggles with Instant Checkout - Whether minimum viable test budgets stay this accessible once conversion bidding is fully rolled out, or whether OpenAI starts nudging spend minimums back up as demand increases ## The verdict This update, taken feature by feature, is unremarkable. Bulk APIs and geo exclusions are not innovation, they're maintenance. But the speed at which OpenAI compressed five months of platform maturation, from a $250K enterprise pilot to a full self-serve, conversion-bidding, bulk-manageable ad platform, is the actual story, and it rhymes with every ad platform that ever mattered. I'm not telling you to move your media budget wholesale. The measurement gaps are real, the $100M figure is self-reported, and EMARKETER's sub-$1B forecast is a useful reality check against the hype. But I am telling you that the advertisers who spend the next two or three months building a small, disciplined test here, clean conversion tracking, modest budgets, LTV:CAC framing instead of raw conversion chasing, will understand this channel better than their competitors when the arbitrage window closes and everyone else shows up at once. That's how it went with AdWords. That's how it went with Facebook. My money says it plays out the same way here :) Wrestling with where ChatGPT Ads fits your own channel mix? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### 70% of Shoppers Now Ask AI First: What That Means for Your Consideration Set URL: https://www.martechnext.com/70-of-shoppers-now-ask-ai-first-what-that-means-for-your-consideration-set/ Last updated: 2026-07-26T08:32:15.000Z **In this article:** - What a new Roland Berger study of 6,000 consumers across nine countries actually found about AI-driven product research - Why "70% use AI to research products" is a bigger deal than it sounds, and where the number needs scrutiny - Why brand reputation is losing ground to recommendations, functionality, and price, and what that means for AI visibility - How this connects to the conversation-as-funnel thesis and the early-mover advantage in GEO - A concrete, five-step plan for getting your brand "seen" favorably by AI research tools, starting now Picture a mid-sized appliance brand's marketing director opening a dashboard next quarter and noticing that organic search traffic is flat, paid search is getting more expensive, and yet product page visits from unfamiliar referrers keep climbing. A little digging shows the traffic is coming from people who asked ChatGPT or Gemini "what's the best dishwasher under 600 euros" and clicked through to whatever got recommended. The brand wasn't mentioned. A competitor was. That scenario is no longer speculative. According to a new study by Roland Berger, based on a survey of 6,000 consumers across nine countries (Germany, France, Spain, Italy, and the Netherlands in Europe; the US, Brazil, and Mexico in the Americas; and the UAE in the Middle East), 70% of 18 to 64-year-olds now regularly use AI tools for product research. Among 18 to 24-year-olds, that figure jumps to 93%. If that holds up, it's the clearest third-party confirmation yet of something I've been arguing for a while: the conversation is becoming the new funnel, and brands that are invisible inside it are quietly falling out of consideration. ## What the study actually found The Roland Berger findings, as reported by Emerce, go beyond a single adoption number. A few data points stand out: - AI tools are increasingly filtering options, comparing products, and shaping recommendations, essentially doing pre-purchase work that used to happen on comparison sites or in search results. - Brand reputation is losing its grip: only 21% of respondents now cite it as a primary purchase driver. - Word of mouth is doing more heavy lifting than brand marketing: over 60% of consumers say they discover products through friends, family, or direct recommendations. - Demonstrable value, functionality, sustainability, and ease of use now outweigh innovation claims or brand power alone. - The market is splitting in two directions: one segment paying for durability and quality, the other chasing discounts. 42% of respondents cut luxury spending in 2025, a third shopped more at discounters, and 23% plan to increase that shift further in 2026. - Dutch consumers are notably more skeptical of AI than the international average (36% versus 24%), with trustworthiness (53%) and privacy/data security (50%) as top concerns. Put together, this isn't just "people are trying ChatGPT." It's a description of a purchase journey where AI does the shortlisting, peer recommendations do the validation, and brand messaging does less work than it used to. That's a structural shift, not a fad. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260726-1024-internationale-studie-ai-verandert-consumentengedrag-visual-1.png) ****Where AI sits in the new purchase journey** โ€” AI tools now filter and recommend before word of mouth validates and brand messaging closes. ยท Data: Roland Berger, 2026 (via Emerce) ## Why the 70% number deserves scrutiny, not dismissal Here's where I have to apply the same standard I'd apply to any vendor claim, including my own: what exactly does "regularly use AI for product research" mean? The Emerce write-up doesn't spell out whether this includes passive touchpoints like Google AI Overviews (which almost every searcher now sees whether they want to or not) alongside active use of ChatGPT, Gemini, or Perplexity for comparison shopping. Those are very different behaviors with very different implications for marketers. Lumping them together can inflate an adoption number that's directionally right but imprecise. It's also self-reported survey data, not platform clickstream data. People are notoriously bad at accurately describing their own digital habits ("I use AI regularly" means different things to a 22-year-old and a 55-year-old). And while the study is refreshingly specific about sample size and country coverage for a report of this kind, six thousand respondents across nine countries in one bucket ("18-64") still means real variance is getting smoothed out. I'd want to see the country-by-country and age-bracket breakdown before quoting this number in a board deck. The bigger gap: the study describes behavior (people are asking AI), not outcome (brands absent from AI recommendations measurably lose sales). That causal link is the one every marketer actually cares about, and it's the one piece of evidence I haven't seen published yet, from Roland Berger or anyone else. Correlation between AI usage and reduced brand loyalty is strongly suggested here. Proof that invisibility in AI answers directly costs market share is still, honestly, a thesis rather than a fact. None of that makes the study wrong. It makes it directionally credible and worth taking seriously, while reserving the specific percentage for "roughly seven in ten," not gospel to three significant figures. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260726-1024-internationale-studie-ai-verandert-consumentengedrag-visual-2.png) ****What still drives purchase decisions** โ€” Word of mouth and demonstrable value now outweigh brand reputation as purchase drivers. ยท Data: Roland Berger, 2026 ## Why this matters even if the number is a bit soft Here's the thing: even if the true figure is 55% instead of 70%, the strategic implication doesn't change. AI-mediated discovery is growing, brand loyalty is weakening, and peer recommendation plus demonstrable value are winning over polish and reputation. That's consistent with what platform-level signals have already been showing (Meta's Business Agents, WhatsApp commerce, the shift toward answer engines instead of ten blue links, [all of which I've covered before](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/)). This is thesis 6 in practice: **the conversation is the new funnel.** When a consumer asks an AI tool to shortlist products, that AI answer *is* the top of the funnel now, whether or not your brand shows up in it. Feed quality, structured data, review depth, and off-site content that AI models can actually parse and trust are becoming as important as your paid search account used to be. Generative Engine Optimization (GEO) is the new SEO, and I've written about the mechanics of that shift in more [detail](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/). It also reinforces thesis 7: **early movers win channels.** Every platform shift, AdWords, Facebook, TikTok, WhatsApp, has rewarded the brands that built the foundation before the channel matured and everyone else showed up to fight over the same few slots. AI answer engines are still in that early window. The brands investing in structured product data, GEO-friendly content, and trustworthy review signals right now are the ones who'll be the "default recommendation" once this behavior fully mainstreams. The ones waiting for definitive proof will be optimizing for a slot that's already taken. There's a secondary read here too, tied to thesis 4, the data flywheel. If AI tools increasingly favor products with rich, structured, trustworthy first-party data (specs, reviews, verified attributes) over products that just have a strong brand story, then the brands with the best data infrastructure have a real edge, [independent of ad spend](https://www.martechnext.com/the-power-of-intelligent-personalization-how-ai-will-transform-the-customer-journey/). ## What the discount and skepticism data adds Two more findings deserve marketer attention, because they change what "winning the consideration set" actually requires. First, price sensitivity looks structural, not cyclical. With over 40% of respondents cutting luxury spend and a third shifting toward discounters, and nearly a quarter planning to lean further into that in 2026, this isn't a temporary reaction to inflation. It's a durable behavior pattern. Combined with brand reputation dropping to 21% as a purchase driver, the message is blunt: showing up in the AI shortlist isn't enough if your actual value proposition (price, durability, functionality) can't survive the comparison once you're there. Second, the Dutch skepticism gap (36% versus a 24% international average) is a useful reminder that GEO and AI-visibility strategy isn't one-size-fits-all across markets. A brand rolling out AI-facing content globally should expect different trust thresholds, and different proof requirements, in different countries. Trust signals (third-party reviews, transparent sourcing, verifiable claims) will matter more in skeptical markets than polished brand copy ever will. ## A five-step starting point You don't need to wait for a perfectly clean causal study to start acting. Here's a practical sequence: 1\. **Audit your AI visibility now.** Ask ChatGPT, Gemini, and Perplexity the questions your customers would actually ask ("best \[category\] for \[use case\] under \[price\]") and see if you show up, and how you're described. 2\. **Fix the data before the marketing.** Structured product data, accurate specs, and up-to-date reviews are what AI models actually pull from. This is the same "right order" principle behind every AI martech project: [data first, automation second, AI last](https://www.martechnext.com/from-rising-acquisition-costs-and-data-chaos-to-ai-driven-growth-in-travel/). 3\. **Invest in third-party proof, not brand copy.** Given that 60%+ of discovery now happens through recommendations and brand reputation alone convinces barely a fifth of buyers, reviews, case studies, and comparison content matter more than another campaign. 4\. **Build for the price-conscious buyer, not just the loyal one.** If durability and demonstrable value are winning over brand story, make sure that's what your product content actually leads with. 5\. **Track off-site mentions as a KPI**, not an afterthought. If AI tools are the new comparison shopping engines, "share of AI recommendation" deserves the same tracking discipline your team already applies to share of search. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260726-1024-internationale-studie-ai-verandert-consumentengedrag-visual-3.png) ****Five-step GEO starting checklist** โ€” A practical sequence for making your brand visible inside AI recommendation environments ## Verdict I'd treat the 70% figure the way I'd treat any single-source survey stat: probably directionally right, not precise to the decimal, and not yet proof of a causal link between AI invisibility and lost sales. But the underlying behavior it describes matches everything else we're seeing across platforms, and the risk of waiting for perfect proof is asymmetric. The brands that treat this as a wake-up call and start fixing their data and off-site presence now will have a real head start. The ones that wait for a definitive McKinsey report to confirm it will be optimizing for a consideration set that's already closed. My take: don't quote "70%" like it's settled science, but don't dismiss it either. Start the GEO audit this quarter. Wrestling with how to make your brand visible (and trusted) inside AI recommendation environments? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Meta AI Can Now Act, Not Just Chat: What That Means for Your Funnel URL: https://www.martechnext.com/meta-ai-can-now-act-not-just-chat-what-that-means-for-your-funnel/ Last updated: 2026-07-25T14:21:22.000Z **In this article:** - What Meta actually announced with Muse Spark 1.1, and why "it acts" is the important word, not "it chats" - Why a mood board for your kitchen renovation is, functionally, a shopping list Meta AI built without you clicking anything - How this extends the thread-as-funnel thesis I've been tracking since [Meta's Business Agents rollout](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/) - The caveats that actually matter: this is a consumer assistant, not (yet) a checkout agent, and the claims are Meta's own - A practical roadmap for getting your brand ready to be recommended, not just ranked --- Picture someone opening the Meta AI app to plan a birthday dinner. They type one sentence: find somewhere for Saturday, check if I'm free, send me a few options. Thirty seconds later they have three restaurant suggestions, cross-referenced against their calendar, ready to book. Nobody searched Google. Nobody opened a reservation app. Nobody scrolled a map with pins on it. The AI did the research, checked the calendar, and shortlisted the outcome. The person just picked one. That is the shift Meta announced on July 24, 2026, and it deserves more attention from marketers than it's getting. Powered by a new model called Muse Spark 1.1, Meta AI can now, in Meta's words, "make plans, work with your apps, and follow through from start to finish." It builds running schedules, scouts Marketplace for furniture that fits a stated budget and style, drafts mood boards, gives daily briefings pulled from your calendar, and keeps working on tasks you only had to set up once. Meta's own framing is blunt: Meta AI "doesn't just think, it acts." That's a marketing line, sure, but it's also an accurate description of a genuine capability jump, from answering questions to executing plans. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/02_Daily-briefing.gif) Meta Copyright ## What Meta actually shipped Strip away the lifestyle examples and three things are new here: 1\. **Multi-step planning.** Meta AI can decompose a vague goal ("help me train for my first half marathon") into a structured plan with recurring check-ins, not a single generated answer. 2\. **App and calendar connectivity.** It reads your schedule, flags conflicts, and adjusts plans around real constraints instead of asking you to do that manually. 3\. **Persistent, recurring execution.** You set a task once (a weekly meal plan, a heads-up on sneaker drops) and it keeps running without re-prompting. That's the "agent" part: it doesn't wait for you to come back. Also notable: everything Meta AI produces (mood boards, training plans, slide decks) lives inside your Meta AI history, which quietly turns the assistant into a persistent workspace, not a search box you visit and leave. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260725-1553-meta-ai-doesn-t-just-think-it-acts-visual-1.png) ****From chat to action: what Muse Spark 1.1 adds** โ€” Four new capabilities turn Meta AI from a chatbot into an execution layer. ยท Data: Meta announcement, July 2026 ## Why marketers should care: the thread just got hands I've written before about the thread becoming the [new funnel](https://www.martechnext.com/why-the-conversation-is-the-new-funnel-not-the-feed/) as commerce moves into messaging and AI answers. Meta's Business Agents already let brands sell inside a WhatsApp or Messenger conversation. What's new here is the other side of that equation: the *consumer's* assistant is now agentic too, not just the *brand's*. That matters because a funnel with two agentic ends behaves very differently from a funnel with one. When Meta AI scouts Marketplace listings for a kitchen renovation and builds a mood board, it is, functionally, running a discovery and consideration phase on the shopper's behalf, unprompted by any brand, invisible to any campaign dashboard, and outside any attribution model built for clicks. The shopper never typed a product name. They typed an intent ("I'm renovating my kitchen, here's my budget and style") and the AI translated that into a shortlist. This is the logical extension of what I called generative engine optimization in the [GEO deep dive](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/): brands don't get found because they rank, they get found because they get *recommended* by a system that's already done the comparison shopping for the user. Muse Spark 1.1 just moves that recommendation step from "answering a question" to "executing a plan," which is a much stickier position to be in. An AI that answers a question once is a search result. An AI that runs your weekly meal plan or tracks sneaker drops for months is a standing subscription to being your shopping assistant, and whichever brands that assistant defaults to win by default, repeatedly, without a single ad served. That's the part worth sitting with: this isn't a new channel to bid on. It's a new decision-maker to be legible to. ## The mood board is a shopping list Here's a concrete, hypothetical example of why this matters operationally, not just philosophically. Say a home goods retailer sells kitchen fixtures through Marketplace. Under the old model, that retailer competes on search terms, images, and price inside a marketplace listing page. Under the new model, when Meta AI is asked to build a renovation mood board, it's making editorial-style judgment calls: which fixtures suit "this style," which fit "this budget," which combination "comes together" well. That's a recommendation engine deciding, on the user's behalf, who gets shortlisted before the user ever compares options themselves. This is the same one-to-one-beats-segments logic I keep coming back to: the constraint that used to force brands into broad segments and generic listings is disappearing, and it's disappearing on the *demand* side now too. The AI isn't matching a segment to an offer. It's matching one person's stated style and budget to one shortlist. Product data quality, structured attributes, and consistent Marketplace listings stop being back-office hygiene and start being the actual battleground, because an AI agent can't recommend what it can't confidently parse. If your product feed is inconsistent across channels (different prices, missing attributes, stale inventory), you're not just risking a bad ad. You're risking exclusion from a shortlist you'll never see being built. ## The caveats that actually matter I'd be doing exactly what I criticize vendors for if I took Meta's own framing at face value here, so two things genuinely matter before anyone reorganizes a Q3 roadmap around this. **First: this is a consumer productivity assistant, not a commerce agent, yet.** The examples Meta shared (kitchen mood boards, marathon training plans, birthday dinner logistics) are personal life admin, not transactional checkout flows. Marketplace browsing is the closest thing to commerce in the announcement, and even there, the AI is surfacing options, not completing purchases. [Meta's Business Agents](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/) are still the more direct commerce play. Muse Spark 1.1 is the layer that will eventually feed intent *into* that commerce layer, but "eventually" is doing real work in that sentence, and Meta hasn't confirmed a timeline for merging the two (confirm from announcement). **Second: the capability claims are Meta's own, from a launch post, with no independent benchmark attached.** "Plans, follows through, course-corrects in real time" is a compelling pitch. It's also exactly the kind of claim that deserves the same scrutiny I'd apply to any startup's demo video. Agentic AI has a well-documented gap between announced capability and reliable, repeated real-world execution. I'd want to see this working consistently across a few hundred real user sessions, not a handful of curated examples, before treating "it acts" as a settled fact rather than a launch promise. Rollout scope (which markets, which languages, which apps it connects to beyond calendar and Marketplace) also isn't fully specified yet (confirm from announcement). Neither caveat is a reason to ignore this. Both are reasons to treat it as an early signal to prepare for, not a channel to buy into this quarter. ## What to do now You don't need a Meta AI campaign strategy yet, because there isn't a buyable unit here. What you need is to be *legible* to an agent that's increasingly making decisions before your brand gets a say. A practical short list: 1\. **Audit your product data as if an AI, not a person, is the first reader.** Structured attributes, consistent pricing across channels, clean Marketplace listings. This is the same discipline [GEO](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/) requires, and it's now doing double duty. 2\. **Get your Business Agent presence in order** if you're on WhatsApp or [Messenger](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/), because the agentic consumer assistant and the agentic brand assistant are on a collision course, and you want to already be in the room when they meet. 3\. **Watch for the merge, not the launch.** The real inflection point isn't "Meta AI can plan a dinner." It's the day Meta AI can plan a dinner *and* book the table through a Business Agent in the same thread. Track that announcement closely; it's the one that turns this from interesting to urgent. 4\. **Treat this as evidence for a pattern, not a one-off.** Every major platform (Google, OpenAI, now Meta) is racing toward agentic assistants that act on the user's behalf. The brands that win the next five years won't be the ones with the biggest ad budgets on these platforms. They'll be the ones whose data is clean enough, and whose offers are structured enough, to be confidently recommended by something that isn't human. ## Verdict Meta AI going agentic isn't a martech tool launch, and I wouldn't treat it as one. It's a signal about where consumer decision-making is heading: away from search-and-compare, toward delegate-and-review. The claims are Meta's own and the commerce piece isn't built yet, so there's nothing to "activate" this week. But the direction is consistent with everything else I've been tracking since Meta's [Business Agents ](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/)rollout: the conversation is becoming the funnel, and now the consumer's side of that conversation has hands, too. My advice: don't wait for a Meta AI ad product to appear before you act. By the time it does, the brands with clean product data and a functioning Business Agent presence will already be the ones getting recommended. Everyone else will be optimizing for a funnel that quietly moved somewhere else. Wrestling with what an agentic assistant layer means for your own funnel? I'd love to exchange thoughts. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260725-1553-meta-ai-doesn-t-just-think-it-acts-visual-2.png) The brand's assistant and the consumer's assistant are converging on the same thread \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### When the Agent Is the Customer: Who Owns the Data When ChatGPT Does the Shopping? URL: https://www.martechnext.com/when-the-agent-is-the-customer-who-owns-the-data-when-chatgpt-does-the-shopping/ Last updated: 2026-07-25T07:53:32.000Z **In this article:** - Why agentic commerce is real in discovery, but still unproven at checkout - The actual risk: not visibility, but losing the interaction that builds your data flywheel - What OpenAI's Instant Checkout retreat and Google's Universal Commerce Protocol tell us - Where the opportunity still sits, and why it's bigger than "getting the AI to recommend you" - A practical way to think about where you want the agent, and where you don't - A clear verdict on how seriously to take this right now Picture a customer who asks an AI agent to buy moisturizer. The agent compares a thousand SKUs on price, reviews, and delivery speed, and picks one. The customer never sees the packaging. Never reads the brand story. Never gets the chance to notice the product they already loved sitting right there in the results. The agent decides, the customer receives a box, and something quiet but important disappears from the process: the moment of choosing. That scene comes from strategist Jess Graham, cited in a sharp piece on MarTech.org called "[The real risk in agentic commerce](https://martech.org/the-real-risk-in-agentic-commerce/?ref=martechnext.com)". It's worth reading in full, because it names something a lot of martech teams have been dancing around: the real threat of agentic commerce isn't visibility. It's ownership. Specifically, ownership of the interaction data that makes your personalization, your retention, and your entire growth engine smarter over time. I've written before about the data [flywheel](https://www.martechnext.com/focus-on-value-not-volume-why-e-commerce-companies-should-care-more-about-their-ltv-cac-ratio/) as one of the core theses of this newsletter: every interaction a customer has with your brand should make your system smarter, and that compounding advantage, not the underlying model, is your real moat. Agentic commerce is the first real test of that thesis at scale. If the agent sits between you and the customer, who's collecting the data that spins the flywheel? You, or OpenAI, or Google, or whoever owns the agent? That's the question this article is built around. Not "how do I rank in ChatGPT," but "what happens to my customer relationship when I'm no longer the one having the conversation." ## Discovery is already agentic. Checkout mostly isn't, yet Let's separate hype from what's actually shipped, because the two protocols making news right now tell two very different stories. OpenAI launched Instant Checkout in September 2025, allowing purchases to complete inside ChatGPT. Five months later, in March 2026, it pulled the feature. Only about a dozen merchants had integrated, usage stayed low, and shoppers who researched inside ChatGPT still preferred to finish the purchase on the retailer's own site (MarTech.org, July 2026). OpenAI didn't abandon commerce, it repositioned: the current model is discovery-first, routing shoppers out to merchant storefronts where the retailer handles payment. Google took the opposite bet. Its Universal Commerce Protocol, announced at NRF in January 2026 with partners including Shopify, Etsy, Wayfair, Target, Walmart, and Visa, is live and expanding: cart support, catalog access, and identity linking are already in use (MarTech.org, July 2026). Read those two moves together and the picture sharpens. Checkout is still being negotiated, and nobody has fully cracked it. Discovery is already agentic, and that's where the ground is shifting under marketers' feet right now. The consumer numbers back that up. Salesforce found 39% of consumers, and 54% of Gen Z, have used generative AI to discover and evaluate products (Salesforce, 2026). Adobe has separately tracked measurable traffic from generative AI tools flowing into retail sites (Adobe, 2026). That's not stated interest, that's people behaving differently already. What's much thinner is evidence of agents actually completing purchases on someone's behalf. Salesforce also found 63% of Gen Z say they're interested in AI agents making purchases for them (Salesforce, 2026), but "interested" is a survey answer, not a transaction log. This is a distinction I'd apply the same skepticism to as any vendor benchmark: stated intent and measured behavior are not the same thing, and conflating them is how roadmaps get built on vibes. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260725-0939-the-real-risk-in-agentic-commerce-visual-1.png) ****Discovery vs. checkout: where agentic commerce actually stands** โ€” Consumer discovery via AI is already real; agentic checkout is still early and unproven. ยท Data: Salesforce, 2026 / MarTech.org, July 2026 ## The real risk isn't ranking, it's disappearing from the moment of choice Here's the sharpest idea in the source piece, and it's worth sitting with: a brand an agent buys is one the algorithm selected. A brand a customer asks for by name is something the brand had to earn. The first is procurement. Nobody falls in love with sourcing. That distinction should reframe how martech teams think about "AI visibility." Being legible to an agent (structured data, clean feeds, a product catalog the model can parse) is necessary. I've written about this before in the context of [GEO](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/): getting found by AI answer engines is the new SEO, and the fundamentals rhyme. But legibility to the machine is not the same as presence with the human. You can be perfectly optimized for agentic discovery and still be a stranger to the person who receives the box. And that's where the data flywheel thesis gets tested for real. If an agent handles discovery, comparison, and eventually purchase, the interaction, the browsing behavior, the hesitation, the comparison shopping, all of it happens inside somebody else's interface. You get the order. You don't get the story of how the customer got there. That story is exactly what your recommendation engine, your segmentation (or better, your one-to-one [personalization](https://www.martechnext.com/the-power-of-intelligent-personalization-how-ai-will-transform-the-customer-journey/)), and your retention playbook are built on. This is the same pattern as every previous platform shift, just one layer deeper. When Facebook's algorithm sat between brands and audiences, marketers lost first-party data too, and spent a decade building owned channels (email, WhatsApp, loyalty apps) to claw the relationship back. Agentic commerce raises the stakes, because now the intermediary isn't just distributing your message, it's making the purchase decision. **Let's be honest:** most marketing teams are not close to ready to answer "what data do we actually own if the agent buys instead of the customer." That's not a criticism, it's a genuinely new question, and 2026 is the year to start answering it deliberately rather than by accident. ## Where the real opportunity is (and it's bigger than SEO for robots) None of this means retreat. The opportunity here is real, and it's bigger than the narrow "how do we get ranked" question most teams are stuck on. First, discovery-layer agents are a new acquisition channel, and early movers win channels. Every platform shift, AdWords, Facebook, TikTok, WhatsApp, has rewarded whoever built the technical and content foundation before the channel matured. Clean product feeds, structured data, and consistent brand information across the web are cheap to build now and expensive to retrofit later. Do that work today. Second, and more important: agentic discovery is exactly the argument for owning the relationship channel that sits outside the agent's reach. If ChatGPT or Google's protocol handles comparison shopping, the moment that still belongs to you is everything after the sale, and everything you can pull the customer back into directly. That's WhatsApp, email, loyalty, a direct app, a preference center, whatever your owned channel is. The brands that win in an agentic world won't be the ones who out-optimize the algorithm. They'll be the ones who give customers a reason to come back and ask for them by name, bypassing the agent entirely next time. Full disclosure, as always: I'm the founder of Zoey, a WhatsApp journey automation platform for the travel industry, so I have skin in this game. But the logic holds regardless of channel: if an intermediary owns discovery, your job is to own retention and re-engagement so completely that the customer starts the next journey with you directly, not through the agent again. Third, this is a genuine argument for investing in first-party data infrastructure now, before agentic checkout matures and the leverage shifts further away from brands. The right order still applies here: [structure your data](https://www.martechnext.com/from-rising-acquisition-costs-and-data-chaos-to-ai-driven-growth-in-travel/), automate the repetitive layer, then apply AI on top. Teams that have that foundation will be able to plug into whatever commerce protocol wins (OpenAI's, Google's, or something neither has announced yet) without scrambling. Teams still sitting on data spaghetti will be locked out of the interesting part of this shift no matter how well they rank. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260725-0939-the-real-risk-in-agentic-commerce-visual-3.png) ****Where the agent sits: discovery vs. relationship** โ€” Agents are taking over discovery; the post-purchase relationship is still yours to defend ## A practical way to think about it I'd break this into three questions every martech or CRM lead should be asking this quarter: 1\. **Where in our funnel does an agent currently intervene, or is likely to soon?** Map discovery, comparison, and purchase separately. Right now the honest answer for most brands is "discovery, maybe, purchase, not yet." That will change, and the protocol wars (OpenAI vs. Google) will decide how fast. 2\. **What data do we lose if that stage goes agentic?** If an agent completes the purchase, you lose the browsing signal, the comparison behavior, the moment of hesitation. That's real flywheel data. Decide now what you can still capture post-purchase (loyalty enrollment, WhatsApp opt-in, a review request) to rebuild the picture. 3\. **Where is our owned channel strong enough that customers bypass the agent and come to us directly?** This is the actual defense. Not better SEO for robots, but a relationship good enough that the customer skips the comparison shopping next time because they already know what they want. None of this requires a grand agentic commerce strategy document. It requires an honest audit of where the agent sits between you and the customer today, and a plan to keep or rebuild the parts of that relationship that matter most. ## My verdict The MarTech.org piece is right to push back on the "how do we get ranked in ChatGPT" framing. That's the easy question, and chasing it alone will leave marketers optimizing for procurement while losing the actual relationship. But I wouldn't overcorrect into alarm either. The data so far (Salesforce, Adobe) shows agents genuinely reshaping discovery. It does not yet show agents reshaping checkout at scale. OpenAI tried and pulled back within five months. Google's protocol is live and growing, but "growing" and "dominant" are different words. So: take agentic discovery seriously, right now, as a channel worth optimizing (clean feeds, structured data, consistency). Take agentic checkout seriously as a coming risk worth preparing for, not a current emergency worth panicking over. And treat owning the post-purchase relationship, the part no agent can currently touch, as the actual moat. The brands that get this right in 2026 won't be the ones who game the algorithm best. They'll be the ones customers still bother to ask for by name. Wrestling with where agentic discovery fits into your own funnel? I'd love to exchange thoughts :) \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### AI Max and the End of the Keyword: What Intent Matching Means for Your Search Budget URL: https://www.martechnext.com/ai-max-and-the-end-of-the-keyword-what-intent-matching-means-for-your-search-budget/ Last updated: 2026-07-24T07:07:52.000Z **In this article:** - Why Google's "billions of new monetizable searches" claim deserves scrutiny, not stenography - What AI Max for Search actually changes under the hood (and what Google won't disclose) - Why the conversion-lift numbers Google has published keep shifting: 14%, then 27%, then 15%, then 27% again - What the AAA and Etsy case studies do, and don't, prove - How this connects to the bigger shift from keywords to conversation as the unit of intent - A practical way to test AI Max without handing over your entire search strategy on faith Picture a performance marketing lead sitting down for a Q1 budget review. Search has been the most predictable line item in the plan for a decade: you know your keywords, you know your match types, you know roughly what a click costs. Then someone forwards a slide from Google's latest earnings call. The company's chief business officer says AI Max is unlocking "billions of net new searches that weren't really monetizable before" (Alphabet Q4 2025 earnings call, via ppc.land). No number. No baseline. No timeframe. Just "billions," repeated across marketing press with the confidence of a settled fact. That's the moment worth pausing on. Not because AI Max is nothing, it's genuinely one of the more important shifts in Search this year, but because the way Google is framing it tells you a lot about where the accountability line is moving in AI-driven advertising. ## What AI Max actually is Let's start with what's real. AI Max for Search isn't a new campaign type, it's an optimization layer you switch on inside your existing Search campaigns with one toggle (Google Ads Help, support.google.com). Once enabled, it does three things: 1\. Expands matching beyond your keyword list using broad match and keywordless signals, so ads can serve on queries you never explicitly targeted. 2\. Generates and tests ad creative and final URLs automatically, including URL expansion to send traffic to whichever landing page Google's system thinks converts best. 3\. Uses AI to interpret the intent behind a search, not just the words in it. That last point is the real story. For twenty years, Search has run on lexical matching: you bid on strings of text, Google matches strings of text, and the gap between "running shoes for flat feet" and "shoes that fix pronation" was your problem to solve with a bigger keyword list. AI Max collapses that gap. It matches on what the searcher means, not what they typed. What Google does not disclose is how. There's no public detail on whether AI Max runs on Gemini, on an embeddings model, or on an evolution of Performance Max's existing keywordless system (TipRanks, tipranks.com). The public language is deliberately qualitative: the system is "predicting what they might need next." For an industry that spent a decade obsessing over Quality Score mechanics, that's a strikingly thin explanation for something that now sits between every advertiser and every query. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260724-0857-google-says-ai-max-unlocks-billions-of-new-monetizable-searc-visual-2.png) ****Keyword matching vs. intent matching** โ€” AI Max shifts Search from matching text strings to matching the meaning behind a query. ## Tracing the "billions" claim Here's the thing about "billions of new monetizable searches": it originates from exactly one place, an earnings call. Alphabet CBO Philipp Schindler said it during the Q4 2025 report, in the context of Alphabet posting $119.8 billion in quarterly revenue, up 24% year over year, with Search ad revenue growing double digits (ppc.land, bestmediainfo.com). Google's own "Think with Google" page repeats the line with a footnote, but the footnote doesn't lead anywhere. No dataset. No definition of "monetizable." No comparison period. That matters for a simple reason: this was said to investors, in a quarter where Google needed a growth story for Search in an AI-answers world. "We found billions of new searches to sell ads against" is exactly the sentence a market wants to hear when the narrative elsewhere is that AI chat interfaces might shrink traditional search volume. I'm not saying the claim is false. I'm saying it's investor relations framing repeated as product fact, and those are different things. If you've read my piece on [the value-over-volume thesis](https://www.martechnext.com/focus-on-value-not-volume-why-e-commerce-companies-should-care-more-about-their-ltv-cac-ratio/), you'll recognize the pattern: a big, round, unfalsifiable number standing in for a metric you could actually act on. ## The conversion numbers that keep moving If the "billions" line is unverifiable, the conversion-lift numbers are at least numbers, but they don't sit still. - At the May 2025 launch, Google's blog cited a **14%** average lift in conversions or conversion value at similar CPA/ROAS, rising to **27%** for advertisers still mostly running exact and phrase match (TipRanks, quoting Google's blog). - On the Q4 2025 earnings call, Schindler cited **15%** more conversions or conversion value for advertisers using AI Max or Performance Max on Search, at similar ROAS (ppc.land). - Google's current "Think with Google" page is back up to **27%**, again framed around advertisers coming from exact/phrase-match-heavy setups. None of these are necessarily wrong. But 14, 15, and 27 are three different numbers, describing what look like three different advertiser cohorts over three different periods, and Google has never reconciled them in one place. If you're the marketer deciding whether to flip the AI Max switch on your biggest campaign, "somewhere between 14% and 27%, depending on who you already were before you started" is not a number you can put in a business case. It's a range that flatters whichever side of it Google wants to lead with that quarter. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260724-0857-google-says-ai-max-unlocks-billions-of-new-monetizable-searc-visual-1.png) ****AI Max's shifting conversion-lift claims** โ€” Google has cited four different conversion-lift figures for AI Max across three separate announcements. ยท Data: TipRanks, ppc.land, Think with Google ## The case studies: real, but not proof Two named examples give the claim some texture. AAA Auto Club Enterprises reportedly saw a 17% improvement in conversion volume using AI Max for personalized creative. Etsy attributed part of a 36% year-over-year GMV increase over the 2025 holiday period, plus a 21% jump in new buyer acquisition, partly to AI Max (both via Google/Think with Google case studies). These are worth taking seriously. Etsy's scale alone makes a 36% GMV swing notable, even shared across multiple causes. But both examples share the same structural weakness as every vendor case study I've ever scrutinized, including, to be fair, the ones I've published myself: Google selected them, Google published them, and there's no disclosed holdout methodology. We don't know what "partially attributed to" means in percentage terms, and we don't know what Etsy or AAA would have seen with a clean control group running the old matching logic in parallel. Directionally useful. Not independently audited. Treat it that way. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260724-0857-google-says-ai-max-unlocks-billions-of-new-monetizable-searc-visual-3.png) ****Case studies: what's disclosed vs. what's missing** โ€” Both case studies report a headline figure without a disclosed control group or attribution methodology. ยท Data: Google / Think with Google case studies ## Why this matters more than a feature update Step back from AI Max specifically and look at the pattern. This is the same move Google made with broad match, then with Performance Max, and it's the same move the whole industry is making around AI-mediated demand capture: the unit of targeting is shifting from the keyword to the intent, and increasingly, to the conversation. [I've written before about how commerce is moving into messaging and AI answer surfaces](https://www.martechnext.com/meta-conversations-2026-the-business-agent-is-here-and-it-changes-your-whatsapp-strategy/), and about how brands need to think about GEO the way they once thought about [SEO](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/). AI Max is the Search-side version of that same shift. Google isn't just getting better at reading your keyword list, it's replacing the keyword list as the primary unit of advertiser control. That's thesis #6 in action: the conversation, or in this case the intent behind the query, is becoming the new funnel, and Search is adapting to stay inside it. That's genuinely exciting if you think about what it unlocks. A travel brand that only bid on "flights to Lisbon" was always leaving money on the table from the traveler who typed "cheapest way to get to Portugal in October with two kids." Intent matching closes gaps human keyword lists never could, the same logic behind why one-to-one recommendation beats static segments (thesis #3). Early movers who learn to work with intent-based matching, rather than fight it, are likely to capture a disproportionate share of that newly reachable demand before competition (and CPCs) catch up, the same pattern we saw with early AdWords, early Facebook, and early WhatsApp adopters (thesis #7). But it's also, honestly, a control question. AI Max is by definition keywordless in parts of its matching, which means less visibility into exactly which queries triggered your ad (Adalysis has written specifically about the reporting gaps this creates, adalysis.com). That's thesis #5, automation with human accountability, tested in real time: Google is asking advertisers to trust the intent-matching layer with less line-of-sight than the keyword era ever required. Whether that trust is earned isn't something Google's case studies can answer. Only your own data can. ## What to actually do about it You don't need to boycott AI Max, and you don't need to flip it on blind, either. Here's the sequence I'd run: 1\. **Turn it on in one campaign, not your whole account.** Pick a mid-size Search campaign with clean conversion tracking, not your top-of-funnel brand campaign, and not something with murky attribution. 2\. **Run it as a genuine holdout test**, not a before/after comparison. Split budget or split by geography/audience so you have a clean control group running your existing keyword strategy in parallel. 3\. **Measure** [**LTV:CAC**](https://www.martechnext.com/from-ltv-cac-thinking-to-ai-driven-growth-how-e-commerce-companies-build-a-sustainable-business-model/) **and incrementality, not conversion count.** A 15% lift in conversions means nothing if those conversions skew toward lower-value customers or convert on branded terms you'd have won anyway. 4\. **Pull search term reports weekly while you test.** Yes, visibility is reduced, but it isn't zero, and the gaps Adalysis and others have documented are worth understanding before you scale spend. 5\. **Set a decay-rate check.** If the lift is real, it should hold up over 60-90 days, not just in the first flush of a new inventory pool that hasn't had time to get competitive (and expensive) yet. 6\. **Treat every published percentage, 14, 15, 27, or otherwise, as directional**, not as your expected outcome. Your account, your audience, and your baseline match-type mix will move the number in either direction. The honest version of AI Max's pitch isn't "billions of new searches, take our word for it." It's "we've built a matching layer that's genuinely better at reading intent than your keyword list ever was, and the way to find out if that's true for you is to test it like you'd test anything else that touches your budget." ## Verdict AI Max is a real capability shift, not a rebrand, and I'd be surprised if intent-based matching doesn't eventually become the default way Search works for everyone. The direction is right. But "billions of new monetizable searches" is a talking point built for an earnings call, not a metric you should plan a quarter around, and the conversion-lift numbers Google has published are inconsistent enough that I wouldn't take any single one at face value. My take: adopt AI Max early, on a contained slice of your account, with a genuine holdout test running alongside it. Early movers who learn the new matching logic while it's still cheap will likely do well out of this. Just don't confuse Google's growth story with your own performance data. Those are, as always, two different things ;) Wrestling with how much control to hand over to AI-driven matching in your own account? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### WhatsApp's iPad Accounts and CarPlay Update Aren't Cute Features, They're Platform Strategy URL: https://www.martechnext.com/whatsapps-ipad-accounts-and-carplay-update-arent-cute-features-theyre-platform-strategy/ Last updated: 2026-07-23T08:21:03.000Z **In this article:** - What Meta actually shipped for WhatsApp this week: standalone iPad accounts, revamped CarPlay/Android Auto, in-chat PDFs, Status music sharing - Why a consumer feature bundle is actually a platform-expansion move, and what that means for the "conversation is the new funnel" thesis - The one feature (in-chat PDFs) that quietly matters more for marketers than the other three combined - Where I'd caution against over-reading a UX drop as a business-platform announcement - A short checklist for what to actually do with this news this quarter Picture a customer support lead at a mid-sized airline. In the morning, she replies to a stranded passenger from her phone. On the drive to the office, she hears the next message read aloud through CarPlay and dictates a reply hands-free. At her desk, she opens a boarding pass PDF a colleague sent, straight inside the chat, annotates it, and sends it back. That evening, from her iPad on the couch, she checks in on the same conversation thread, no phone required. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/01_WA-New-Feature-Roundup_Carousel-02.jpg) Nothing in that day involved a new marketing feature. But it shows exactly what Meta just built: a version of WhatsApp that follows you across every screen and every moment, not just the one in your pocket. That's the story worth paying attention to, more than any single feature on the list. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/01_WA-New-Feature-Roundup_Carousel-01.jpg) ## The details On July 22, 2026, Meta announced a set of WhatsApp updates rolling out across devices (Meta Newsroom). The headline items: - **Standalone iPad accounts.** You can now register a WhatsApp account directly on iPad, no phone-linking required. - **Revamped CarPlay and Android Auto.** A refreshed hands-free audio experience: hear and respond to messages, make calls, view call history, and reach favorites from the car's screen. - **In-chat PDFs.** PDFs now open directly inside WhatsApp on web and desktop, with lightweight annotation and highlighting, powered by Adobe Acrobat. - **Status music sharing.** Users can share a song from Apple Music or Spotify straight to their WhatsApp Status. Meta framed it simply: *"WhatsApp can keep up with how you actually live"* (Meta Newsroom). All four features are rolling out now. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260723-1004-new-whatsapp-features-keep-up-with-the-way-you-actually-live-visual-2.png) From single-phone app to an all-day, all-device environment. ## The MartechNext take ### This is a consumer release, but it's evidence for a marketing thesis None of these four features is a business tool. There's no new template type, no new API endpoint, no pricing change. If you're scanning this for a reason to rebuild your WhatsApp Business flows tomorrow, there isn't one, yet. But step back and the pattern is clear. Meta isn't adding features to WhatsApp so much as removing the boundaries around it: no longer "the app on your phone," but an environment that follows you into the car, onto a second device, into your working documents, into what you're listening to. That's a deliberate platform play, and it's consistent with something I've been arguing for a while: [**the conversation is becoming the new funnel**](https://www.martechnext.com/meta-conversations-2026-the-business-agent-is-here-and-it-changes-your-whatsapp-strategy/). Every time Meta widens the number of daily moments where a WhatsApp thread is open, it widens the surface where a brand conversation, a support flow, or eventually a purchase can happen too. This also fits the pattern I flagged with [Meta Business Agents](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/): early movers on a platform shift capture disproportionate advantage, because habits and defaults get set before the channel fully matures. WhatsApp becoming an all-day, all-device presence for consumers is exactly the kind of shift that rewards brands who are already fluent in the channel when the commerce layer catches up to the consumer layer. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260723-1004-new-whatsapp-features-keep-up-with-the-way-you-actually-live-visual-3.png) ****Four updates, one platform tell** โ€” Each July 2026 feature and what it actually signals for marketers. ยท Data: Meta Newsroom, July 2026 ### The feature that actually matters: PDFs living inside the thread Of the four updates, the one marketers should sit with is in-chat PDFs. Right now, a lot of WhatsApp business communication still bounces the customer out of the thread: a booking confirmation link, an invoice attachment that opens a separate viewer, a contract that gets emailed "for your records." Every one of those exits is a small tax on the relationship, a moment where the customer leaves the conversation to do something the conversation should have handled natively. Adobe Acrobat-powered PDFs that open and can be annotated directly inside WhatsApp point toward a future where the itinerary, the invoice, the signed agreement, doesn't just get *sent* through the thread, it *lives* in the thread. That's a small technical detail with a real business implication: fewer handoffs, fewer chances to lose the customer to email, more of the relationship staying inside a single, ownable channel. Say a mid-sized tour operator sends a 12-page itinerary as a PDF today. The customer downloads it, maybe forwards it to a travel companion by email, and the WhatsApp thread goes quiet until the next automated update. If that document instead lives natively in the chat, with the ability to annotate ("confirmed, changing the Tuesday activity") right there, the thread stays the single source of truth for the whole trip. That's not a huge leap in imagination, it's just following where Meta is clearly pointing. ### CarPlay and Android Auto: attention, not conversion, and that's fine The revamped in-car experience is genuinely useful for consumers, and it's a reminder that "read rate" is going to mean something different soon. Hands-free, voice-read WhatsApp messages mean your business notifications, delivery updates, appointment reminders, boarding alerts, might get consumed by ear while someone's driving, not read on a screen. That has a practical consequence for anyone writing WhatsApp Business templates: messages need to work when heard, not just when read. A message stuffed with emoji, links, and abbreviations reads fine on a screen and sounds like nonsense out loud. Nobody's measuring this yet, and I wouldn't rewrite your entire template library over one feature announcement, but it's worth testing a handful of your highest-volume automated messages by having someone read them aloud. If it sounds like it was written for a screen, it probably was. ### Status music sharing: minor feature, bigger inventory story On its own, sharing a song to your Status is not a marketing event. But it's one more data point in Meta's ongoing investment in Status as a content format, richer Status content historically precedes richer Status monetization. I've flagged [Status Ads as a channel worth watching](https://www.martechnext.com/whatsapp-marketing-in-2026-what-marketers-need-to-know-to-use-it-effectively/), and every feature that makes Status a more expressive, more frequently used format increases the value of that inventory down the line. Worth tracking, not worth building a campaign around yet. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260723-1004-new-whatsapp-features-keep-up-with-the-way-you-actually-live-visual-1-1.png) Four July 2026 WhatsApp updates, ranked by near-term marketing relevance. ยท Data: Meta Newsroom, July 2026 ## Where I'd push back on over-reading this Two honest caveats, because a feature bundle is not a strategy shift and shouldn't be treated as one. First, **nothing here has been confirmed for the Business Platform or Cloud API**. In-chat PDF annotation is currently described for web and desktop, for personal use (confirm from announcement whether or how this extends to Business API document delivery). If your team is planning to send annotatable documents through automated business flows, check Meta's Business Platform documentation directly before you commit a roadmap to it. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/01_WA-New-Feature-Roundup_Carousel-03.jpg) Second, don't confuse *reach* with *response*. A message read aloud through CarPlay is a message that got heard, that's attention, not action. It's tempting to fold "voice-read in the car" into your engagement metrics as if it behaves like an opened message on a screen, but until Meta or independent research clarifies how (or whether) these interactions are measured and attributed, treat it as a channel-consumption statistic, not a conversion statistic. That's the same discipline I'd apply to any channel [open-rate claim](https://www.martechnext.com/focus-on-value-not-volume-why-e-commerce-companies-should-care-more-about-their-ltv-cac-ratio/). ## What to actually do with this, this quarter 1\. **Read your top five automated WhatsApp templates out loud.** If they don't work as spoken sentences, rewrite them. CarPlay and Android Auto make this a real, if small, consideration now. 2\. **Map your document-heavy customer touchpoints** (itineraries, invoices, contracts, boarding passes) and flag which ones currently force the customer to leave WhatsApp. That's your shortlist for when native, in-chat document handling reaches the Business API. 3\. **Keep an eye on Status**, not because this update changes anything for advertisers today, but because Meta is clearly still investing in the format. If you've been testing Status Ads, or considering it, this is a "watch, don't wait" signal. 4\. **Don't retrofit a strategy document around a consumer feature drop.** Confirm anything business-critical against Meta's actual Business Platform changelog before you brief your team. 5\. **Revisit your WhatsApp channel mix conversation** with this frame: WhatsApp is no longer "the messaging app," it's becoming the connective layer across a customer's devices and moments. That's a bigger claim on attention than most brands are currently designing for. None of this requires a rebuild by next week. But if you're still treating WhatsApp as a single-device notification channel, this update is a nudge that Meta isn't building it that way anymore, and your customers are already living in the multi-device version whether your martech stack has caught up or not. **Verdict:** this is a minor release with a major tell. The individual features won't move a single KPI on their own, and I wouldn't spend budget reacting to them directly. But the direction, WhatsApp as an all-day, all-device environment rather than a phone app, is exactly the trajectory the "conversation is the new funnel" thesis predicts, and it's worth building your 2026 channel roadmap with that trajectory in mind, not against it. Wrestling with where WhatsApp actually fits in your channel mix, beyond the notification layer? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Pre-Seed, Big Thesis: Why Investors Are Betting on AI-Search Visibility Tools Now URL: https://www.martechnext.com/pre-seed-big-thesis-why-investors-are-betting-on-ai-search-visibility-tools-now/ Last updated: 2026-07-22T13:09:22.000Z **In this article:** - Why a $2M Toronto pre-seed round is worth your attention even though the number itself is small - What ChatFeatured actually does, and what its funding announcement leaves unsaid - How this round fits a wider wave of capital chasing "AI search visibility" (Profound, Peec AI, and others) - Why this connects to two MartechNext theses: the conversation as the new funnel, and early movers winning channels - The real question every buyer should ask before adopting a GEO tool: is this a moat, or a feature Semrush ships next quarter? - A practical read on whether to pilot, wait, or ignore this category right now --- Picture a mid-sized B2B software company whose marketing director just typed their own product category into ChatGPT and watched three competitors get named before their brand did. No warning, no ranking report, no dashboard alert. Just silence where visibility used to be. That scene is becoming routine, and it's exactly the gap a wave of new startups wants to fill. The latest is ChatFeatured, a five-person Toronto company that just closed an oversubscribed $2 million USD (roughly $2.8 million CAD) pre-seed round, led by Storytime Capital with participation from Garage Capital and BY Venture Partners, according to the company's announcement covered by [BetaKit](https://betakit.com/chatfeatured-secures-2-million-usd-to-optimize-your-companys-chatbot-reach/?ref=martechnext.com) and syndicated via [Morningstar/BusinessWire](https://www.morningstar.com/news/business-wire/20260722829210/chatfeatured-raises-us2-million-to-build-an-autonomous-ai-search-agent-for-marketing-teams?ref=martechnext.com). The company reportedly targeted $1.5 million and raised more on investor demand. Two million dollars is not, on its own, news. What makes it worth a proper look is where it lands: a category where the leader just raised at a reported $1 billion valuation. That gap between a five-person seed-stage team and a unicorn tells you something about how early we are, and how fast investors think this is going to move. ## The details Here's what's confirmed. ChatFeatured launched in the first week of January 2026 out of Toronto and, per the company, already serves customers in five countries. The founding team is Farris Nasr (CEO), Nithiiyan Skhanthan (CTO), and Ryan Nahas (founding Head of Revenue), backed by a five-person team total. The funds are earmarked for product development and go-to-market. The product itself is positioned as an AI search optimization platform: it identifies how a company gets mentioned (or doesn't) by major chatbots and generates a strategy to increase visibility. One line from the announcement sums up the ambition: it aims to build "an autonomous AI search agent for marketing teams." No valuation was disclosed, in either the press release or the BetaKit coverage. No round structure (SAFE versus priced equity) was published. Worth flagging up front, because it matters for how you read everything that follows: I couldn't verify the round size or investor list independently through a third-party filing, so I'm relying on the company's own disclosure, same as everyone else covering this story. ## What ChatFeatured is, and isn't, yet Let's be honest about what a seed-stage funding announcement actually tells you: almost nothing about whether the product works. It tells you a small group of angel and pre-seed investors liked the team, the market timing, and the pitch. That's it. No published customer case studies. No before-and-after visibility metrics. No technical detail on what "autonomous agent" actually means under the hood: which LLMs it monitors, how it scores mentions, whether the "strategy" it generates is genuinely tailored or a templated content plan. I looked for this and came up empty, which isn't unusual at this stage, but it's a gap you should notice before you take the marketing language at face value. None of that makes ChatFeatured a bad bet. Plenty of category-defining products started with less than this. It just means the honest label for where this sits is "promising, unproven," not "market-validated." ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260722-1455-chatfeatured-raises-us-2-million-to-build-an-autonomous-ai-s-visual-1.png) What's confirmed publicly versus what remains undisclosed about ChatFeatured's raise. ยท Data: BetaKit / Morningstar-BusinessWire, 2026 ## The GEO gold rush ChatFeatured is not raising in a vacuum. It's entering a category that's suddenly attracting real capital. Profound, widely seen as the category leader in this space, raised a $96 million Series C at a reported $1 billion valuation in February 2026, according to [Fortune](https://fortune.com/2026/02/24/exclusive-as-ai-threatens-search-profound-raises-96-million-to-help-brands-stay-visible/?ref=martechnext.com). Peec AI, a similar player, raised a $21 million Series A in November 2025 to help brands "win in AI search," per [TechCrunch](https://techcrunch.com/2025/11/17/as-consumers-ditch-google-for-chatgpt-peec-ai-raises-21m-to-help-brands-adapt/?ref=martechnext.com). Line those three up and you get a rough shape of the category: an established leader nearing unicorn status, a well-funded Series A challenger, and a scrappy pre-seed entrant trying to find its wedge. That's a classic early-market pattern, and it usually means one of two things is about to happen: either the category consolidates around two or three winners within eighteen months, or it turns out to be a feature that gets absorbed into tools marketers already pay for. More on that in a minute. I'll flag the obvious limit here too: I have not done a feature-by-feature comparison of ChatFeatured against Profound, Peec AI, Athena, Rankscale, or Otterly.ai. What I can tell you with confidence is that all of them are chasing the same underlying shift, and that shift is the real story. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260722-1455-chatfeatured-raises-us-2-million-to-build-an-autonomous-ai-s-visual-3.png) ****Scale gap: five-person team vs. reported unicorn** โ€” Profound's reported $1B valuation dwarfs ChatFeatured's $2M pre-seed, underscoring how early the GEO category still is. ยท Data: Fortune; BetaKit, 2026 ## Why this matters: the conversation is the new funnel Here's the thesis I keep coming back to on MartechNext: commerce and discovery are moving into messaging threads and AI answers. ChatGPT, Meta's Business Agents, WhatsApp business flows: the funnel doesn't start on a search results page anymore, it starts inside a [conversation](https://www.martechnext.com/meta-conversations-2026-the-business-agent-is-here-and-it-changes-your-whatsapp-strategy/). When a prospective customer asks an AI assistant "what's the best project management tool for a 20-person agency," the answer that gets surfaced is the new page-one ranking. If your brand isn't in that answer, you don't get a click you can retarget. You just don't exist for that moment. That's why GEO (generative engine optimization) is getting funded like a discipline, not a feature. It's SEO's structural cousin: same underlying question (how do I get found), completely different mechanics (you're optimizing for what an LLM says about you, not what a crawler indexes). Feed quality, off-site consistency, and how your brand shows up across scattered third-party content all start to matter in ways classic SEO tooling never measured. And this is where the second thesis kicks in: early movers win channels. Every platform shift, AdWords in the 2000s, Facebook ads in the early 2010s, TikTok organic in 2020, WhatsApp business messaging more recently, rewarded the brands and vendors who built real capability before the channel matured and got crowded. The investors backing ChatFeatured, Profound, and Peec AI are making a bet that GEO is at that early stage right now, and that whoever builds category position and customer habits first has a durable edge later. That's not a guaranteed outcome. But it's a coherent bet, and it's the same bet that made early AdWords agencies and early TikTok-native brands rich. ## The moat question Here's the part I'd push hardest on if I were writing the check: what's the actual moat? Every AI startup claims a data moat. The honest test is asking why an incumbent with distribution couldn't just copy it. Semrush, Ahrefs, and HubSpot already own the SEO and content workflows most marketing teams live in daily. Adding an "AI mention tracker" panel to an existing dashboard is not a hard engineering problem for a company with that user base and that [data pipeline](https://www.martechnext.com/the-power-of-intelligent-personalization-how-ai-will-transform-the-customer-journey/). So the real question for ChatFeatured, and for the category generally, is whether visibility tracking and optimization strategy stays a standalone product worth paying for separately, or becomes table stakes bundled into tools marketers already have open. My honest read: the tracking and reporting layer commoditizes fast. What might stay defensible is the strategic layer: understanding *why* an LLM mentions one brand over another, and being able to act on that with genuinely tailored content and structured data, not generic advice. That's a harder problem, and it's the one worth watching whether ChatFeatured, or anyone in this category, actually solves. There's also a quieter question worth asking, tied to a thesis I return to often: does a tool like this help a brand get its product and content data in proper shape before chasing AI mentions, or does it jump straight to "AI-generated strategy" on top of the same messy, inconsistent brand data that's always been there? MIT's widely cited research puts AI project failure rates at 95%, largely because tools get bolted onto [disorganized data rather than structured processes](https://www.martechnext.com/from-rising-acquisition-costs-and-data-chaos-to-ai-driven-growth-in-travel/). I don't know yet which category ChatFeatured falls into. Neither does anyone outside the company, based on what's public. ## The skeptic's lens, proportionately applied A few things worth holding onto before you get too excited or too dismissive: "Oversubscribed" and "following investor demand" are the company's own framing, repeated across a press release and its syndicated pickup, not independently verified by a third party. That doesn't make it false, it just means treat it as marketing copy, not confirmed fact. No valuation, no round structure, no published case studies. That's normal for a five-person, six-month-old company. It's also a reason to withhold judgment on whether this is "smart bet" or "hype" until there's actual evidence: repeat customers, published visibility improvements, a Series A. And the scale gap matters. $2 million next to Profound's $96 million and reported $1 billion valuation isn't a competitive rivalry yet, it's an early entrant testing whether there's room for more than one winner in a category that's maybe eighteen months old. That's worth remembering the next time a GEO vendor's pitch deck compares itself to "the Profound of X." ## What to watch If you're a martech buyer or a founder circling this space, here's what would actually move my read on ChatFeatured, or any GEO tool, from "interesting bet" to "worth adopting": 1\. **A published case study with real numbers.** Not "increased visibility," but specific query sets, specific competitor comparisons, before-and-after mention rates. 2\. **A follow-on round within 12-18 months.** Pre-seed to seed within that window usually signals genuine traction, not just a good pitch. 3\. **Whether Semrush, HubSpot, or a similar incumbent ships a comparable feature.** If they do within the next year, that tells you the standalone-category thesis is shakier than the funding numbers suggest. 4\. **Whether the product actually helps structure brand and content data**, or just layers AI-generated advice on top of whatever's already there. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260722-1455-chatfeatured-raises-us-2-million-to-build-an-autonomous-ai-s-visual-2.png) Three GEO funding rounds, from pre-seed to reported unicorn valuation, in under a year. ยท Data: BetaKit, TechCrunch, Fortune, 2025-2026 ## The verdict I don't think ChatFeatured is a bad bet, and I don't think it's proven either. It's a small, credible team making a coherent wager on a real shift: AI answers are becoming a discovery surface, and almost nobody outside SEO specialists is tracking how their brand shows up there yet. That's a genuine gap, and being early to it is exactly the kind of move that's paid off in every prior platform shift I've watched. What I'd tell a marketing team weighing whether to act right now: don't wait for the category to fully sort itself out before you start monitoring your own AI-search presence, because by the time it's obvious, the cheap seats will be gone. But also don't treat a $2 million pre-seed announcement as proof of anything beyond "smart investors think this space is worth a bet." Pilot, ask for real numbers before committing budget, and watch whether the incumbents move. I don't have all the answers here, and I suspect the category itself doesn't yet either ;) Wrestling with GEO strategy in your own organization? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Google ships three new Gemini models, but the one martech vendors actually need is still missing URL: https://www.martechnext.com/google-ships-three-new-gemini-models-but-the-one-martech-vendors-actually-need-is-still-missing/ Last updated: 2026-07-22T08:35:38.000Z Here's the news that matters more than it looks like it does: Google shipped three new Gemini models this week, and the one everyone building serious AI products actually wanted wasn't among them. If you run a martech stack, or you're evaluating vendors who build their agents on top of Google's models, this is worth five minutes of your attention. Not because Gemini is broken. Because it tells you something concrete about how to think about the AI vendors in your funnel right now. ## The details On Tuesday, Google DeepMind released Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber, according to TechCrunch. 3.6 Flash is positioned as the new workhorse: better coding and multimodal performance, and up to 17% fewer tokens burned per task than its predecessor, which makes it cheaper to run at scale. Flash-Lite is the budget option. Flash Cyber is a specialized security model, limited for now to governments and select partners. What didn't ship: Gemini 3.5 Pro, the flagship reasoning model that Google's own team teased back in May, saying it was "already being used internally" and would roll out "next month." That was two months ago. Bloomberg reported last week that Google is hitting internal delays because 3.5 Pro isn't meeting its own performance targets. Meanwhile Pro hasn't been updated since February, and in that same window OpenAI has pushed out GPT-5.5 and started rolling GPT-5.6, and Anthropic has shipped Opus 4.8, Sonnet 5, and wider access to Fable 5. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260722-1026-google-releases-three-new-gemini-models-but-no-3-5-pro-visual-2.png) While Gemini Pro sat still since February, OpenAI and Anthropic kept shipping Google DeepMind's product lead Logan Kilpatrick said the team is testing 3.5 Pro with partners now and hopes to "land soon." He also confirmed Google has kicked off pre-training for Gemini 4. ## The MartechNext take Let's separate two things, because vendors and marketers keep conflating them: the capability curve and the control curve. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260722-1026-google-releases-three-new-gemini-models-but-no-3-5-pro-visual-3.png) The capability curve races ahead while the control curve โ€” trust, oversight, vendor readiness โ€” moves deliberately slower. The capability curve is what these models can technically do. It's moving fast, arguably faster than anyone's ability to responsibly deploy it. The control curve is a different question entirely: how much trust and oversight does a business actually put around a model before letting it touch customers, budgets, or brand voice? That curve moves much slower, and it should. This is a thesis I keep coming back to, because every quarter gives a [fresh example of the gap](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/). This week's release is a textbook version of that gap. Google didn't hold back Flash models, the ones doing the unglamorous, high-volume work: classification, summarization, quick agent actions. It held back Pro, the model you'd trust with complex reasoning, nuanced copy, or decisions with real consequences if it gets something wrong. That's not an accident. That's Google being honest, in its own quiet way, that Pro isn't ready for the trust level its own roadmap promised. Here's why that's actually useful for you, not just a caveat. If you're evaluating a martech vendor right now, whether it's a personalization engine, a content agent, or a [conversational commerce tool](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/), ask them one direct question: which Gemini model is this built on, and what happens to my output quality if Google swaps it under the hood next quarter? Vendors building on Flash-tier models today are making a defensible bet: lower cost, faster inference, good enough for high-volume operational tasks like tagging, routing, or first-draft generation. Vendors quietly waiting on 3.5 Pro to ship before they finalize their reasoning-heavy features are also making a defensible bet: they'd rather ship something reliable a quarter late than something clever and unstable on time. Both are legitimate strategies. What's not legitimate is a vendor pitching you Pro-level reasoning claims on a Flash-tier foundation, or worse, not being able to tell you which model powers which part of their product at all. The honest caveat here: model swaps under the hood are mostly invisible to you as a buyer, and that's the actual risk, not "Google is falling behind." Whether Google ships 3.5 Pro next month or next quarter says very little about the long-term model race (this is a company mid-way through training Gemini 4, not a company in retreat). It says a lot about whether the vendor sitting between you and that model has a plan for exactly this kind of delay. Ask them. If they don't have an answer, that's the finding, not the six-week gap between Pro announcements. The other thing worth naming: this is the same [commoditization pattern I wrote about with Kimi K3](https://www.martechnext.com/kimi-k3-the-first-open-3-trillion-class-model-and-why-marketers-should-pay-attention/). The model layer keeps getting cheaper and more interchangeable. The moat was never going to be "which frontier model do you use." It's going to be what you build around it: the data flywheel, the workflow, the human judgment layer that decides when to trust the output and when not to. Models come and go on a release schedule. Your customer data and your retention numbers don't. ## Verdict This isn't a red flag for Gemini, and it isn't a green flag either. It's a prompt to do the diligence you should already be doing on every AI vendor in your stack: which model is under the hood, what's the fallback plan if that model changes or slips, and does the vendor's trust level in their own product match the trust level they're asking from you. Google being cautious about shipping Pro is, if anything, a decent sign that somebody at Google is still respecting the control curve. I'd rather work with vendors who show the same restraint than ones who ship first and figure out the trust question later :) \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Your Match Rate is the most important marketing metric youโ€™re probably ignoring URL: https://www.martechnext.com/your-match-rate-is-the-most-important-marketing-metric-youre-probably-ignoring/ Last updated: 2026-07-21T13:13:32.000Z **In this article:** - What "match rate" actually measures, and why every vendor defines it slightly differently - Why the gap between the audience you build and the audience you reach is structurally widening - How a low match rate quietly inflates your CAC and hides your real LTV - Who's selling the fix (Rokt mParticle, LiveRamp, TransUnion/Neustar) and what to actually verify before you buy - Why this is a data-structuring problem before it's ever an AI problem - A short, practical checklist to find out where you actually stand Picture a mid-sized retailer's growth marketer building a lookalike audience from 100,000 loyal customers and pushing it to a major ad platform for a retention campaign. The campaign dashboard looks healthy: decent CTR, reasonable CPM, ROAS in the range leadership expects. Nobody in the review meeting asks how many of those 100,000 people the platform could actually recognize. If the answer is 55,000, the campaign never touched the other 45,000 at all, and nothing in the reporting says so. The numbers just look like a smaller, quieter version of success. That invisible gap has a name: match rate. It's the share of an uploaded audience that a platform can resolve to one of its own known identities, usually by matching hashed emails or phone numbers against its logged-in user base. And according to Rokt mParticle, most performance marketers don't track it, because most performance marketers don't know it's a number at all ([MarTech.org](https://martech.org/why-your-match-rate-is-the-most-important-number-youre-not-tracking/?ref=martechnext.com)) I think that's the most useful thing published on identity in months, and it lands at exactly the moment it should. Third-party cookies are dying, Apple's App Tracking Transparency has cut deep into device-level tracking, and every walled garden keeps tightening how it lets outside data in. In that environment, match rate isn't a niche ad-ops metric. It's becoming the ceiling on everything you do with customer data, and the flip side of that ceiling is an opportunity: the brands with clean, well-resolved first-party identity are about to have a real, compounding edge over everyone still uploading messy lists and hoping for the best. ## What match rate actually measures (and why nobody agrees on the definition) Strip away the vendor language and match rate is simple: you upload a list of customer identifiers, the receiving platform tries to match each one to an identity it already recognizes, and match rate is the percentage that resolves. The mechanics matter more than the definition, though. When you push a first-party audience to Meta, Google, or any ad platform, you're not targeting "your customers." You're targeting the subset of your list the platform can resolve to its own logged-in accounts. Everything else falls out silently, no error, no warning, just a smaller campaign running under the name of a bigger one ([MarTech.org](https://martech.org/why-your-match-rate-is-the-most-important-number-youre-not-tracking/?ref=martechnext.com)). Here's the illustration [Rokt](https://www.mparticle.com/?ref=martechnext.com) mParticle itself uses, and it's a clean one: build an audience of 100,000 customers, upload it somewhere that matches 55%, and your campaign runs against 55,000 people while the other 45,000 simply don't exist to the platform, regardless of how sharp your creative or bidding is. Now here's the part that should make you pause before you take any vendor's match rate number at face value: there is no single, industry-standard definition of "match rate." LiveRamp measures match rate in the context of onboarding a file into its identity graph. mParticle (now Rokt mParticle) measures it in the context of activating an audience against an ad platform's own logged-in base. TransUnion, having folded in Neustar, measures it inside its TruAudience identity graph for "targeting precision." These are related concepts, but they're not the same calculation, and a vendor quoting an impressive match rate percentage is very often answering a slightly different question than the one you're asking (research on this remains incomplete; treat cross-vendor comparisons as directional, not apples-to-apples, until you've confirmed methodology directly with each vendor). ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260721-1436-why-your-match-rate-is-the-most-important-number-you-re-not--visual-1.png) At a 55% match rate, an uploaded 100,000-person audience runs as a 55,000-person campaign, with the rest invisible to the platform. ยท Data: MarTech.org / Rokt mParticle That inconsistency is a genuinely fair reason to be skeptical of any single benchmark you're handed. It is not a reason to ignore the metric. Ask your own platforms, right now, what their match rate methodology is and how they calculate it for your account. Most marketers have never asked the question, which is exactly Rokt mParticle's point. ## Why the gap is widening, not narrowing Three structural forces are driving match rates down across the industry, and none of them are temporary. Third-party cookie deprecation removed the connective tissue that let platforms bridge an identity across sites without you doing anything. Apple's App Tracking Transparency cut off the device identifiers that used to make cross-app matching trivial. And the walled gardens (Meta, Google, Amazon, TikTok) keep narrowing what outside data they'll accept and how they'll resolve it, because tighter matching logic serves their own privacy posture and their own data advantage at the same time. Layer on the mundane stuff that never goes away: the customer who signs up with a work email and buys with a personal one, the phone number formatted three different ways across three different systems, the record that's been stale for two years. None of that is new. What's new is that these ordinary data-hygiene failures used to be quietly compensated for by third-party identity graphs stitching things together in the background. That compensation layer is gone. The stitching now has to happen inside your own systems, or it doesn't happen at all. IAB research, cited by MarTech.org, put a striking number on industry readiness for this shift: 88% of the industry reportedly still faces uncertainty from Google's cookie policy changes ([MarTech.org](https://martech.org/googles-cookie-pivot-leaves-88-of-industry-in-uncertainty-iab/?ref=martechnext.com)). I'd treat that figure as a single, industry-survey data point rather than gospel (I couldn't independently verify the underlying IAB methodology for this piece), but directionally it matches what every martech operator already feels: most teams have not actually rebuilt their identity infrastructure for a world without third-party cookies. They've just watched their match rates quietly erode and adjusted their expectations downward instead. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260721-1436-why-your-match-rate-is-the-most-important-number-you-re-not--visual-3--1-.png) ****Three forces shrinking match rate** โ€” Cookie deprecation, ATT, and tightening walled gardens are compounding, structural pressures, not one-off shocks. ## The business cost: your CAC is worse than you think, and your LTV is invisible This is where match rate stops being an ad-ops curiosity and becomes a straight line to the metric I keep coming back to on this blog: [**LTV:CAC** "Focus on value, not volume".](https://www.martechnext.com/focus-on-value-not-volume-why-e-commerce-companies-should-care-more-about-their-ltv-cac-ratio/) A low match rate inflates your effective CAC in a way that never shows up as a line item. You're paying full price to reach an audience, but the platform is only delivering a fraction of it, and worse, you have no idea which fraction. Maybe it skews toward your most engaged, easiest-to-match customers (the ones with clean, recent, single-device data) and away from the higher-value but messier segments you actually built the campaign to reach. Per-audience-member cost goes up. Nobody notices, because the dashboard reports performance against the matched audience, not the audience you paid to build. The LTV side is quieter and arguably worse. If you can't resolve a customer's identity across channels and touchpoints, you can't see their full value. A customer who books directly on your site, opens your emails on a different address, and interacts with your brand on WhatsApp under yet another identifier looks, on paper, like three partial customers instead of one valuable one. Any retention model, recommendation engine, or churn score you build on top of that fragmented picture is built on top of "data spaghetti," which is exactly the failure mode MIT points to when it explains [why the overwhelming majority of enterprise AI initiatives fail to deliver value](https://www.martechnext.com/from-rising-acquisition-costs-and-data-chaos-to-ai-driven-growth-in-travel/). This is thesis #1 in miniature: value over volume breaks down the moment you can't even reliably count your customers. ## Who's selling the fix, and what to actually check before you buy Rokt mParticle, formed out of the roughly $300 million merger between Rokt and mParticle, has built its "Match Boost" feature specifically to close this gap for audience activation across major ad platforms. It's not the only identity vendor making this pitch. LiveRamp has built its business on identity resolution and match rate as a core product metric for years. TransUnion, after acquiring Neustar, folded that identity data into its TruAudience graph and markets the combination as improved "targeting precision". Google's PAIR and Meta's Advanced Matching offer comparable ideas natively inside the walled gardens themselves. I'll say this plainly: I could not independently verify any specific match rate lift percentage from any of these vendors for this piece, including Match Boost's. That's not a knock on the product, it's a statement about what publicly available reporting currently supports. Vendor-run benchmarks are directionally credible (these companies aren't going to build a whole feature around a made-up problem), but a lift number quoted in a launch announcement is not the same thing as an independently audited result on your specific customer file. If a vendor gives you a match rate improvement claim, ask three follow-up questions before you sign anything: matched against what baseline, measured by whose definition, and verified by whom. That's not cynicism for its own sake. It's the same scrutiny I'd apply to any personalization case study, including ones I've published myself: the claim is usually real, but "real" and "verified on your data" are two different things. ## The MartechNext take: this is a data problem before it's an AI problem Here's why I think this article deserves more attention than it's getting. Match rate is a clean, concrete illustration of a thesis I keep repeating on this blog: [**the right order is data first, then automation, then AI**. ](https://www.martechnext.com/from-rising-acquisition-costs-and-data-chaos-to-ai-driven-growth-in-travel/)You cannot personalize, recommend, or model retention for a customer your system can't reliably recognize across touchpoints. A brilliant recommendation engine trained on fragmented identity data will still produce mediocre recommendations, not because the model is weak, but because the [data feeding it is spaghetti](https://www.martechnext.com/the-power-of-intelligent-personalization-how-ai-will-transform-the-customer-journey/). It's also a direct hit on **the data flywheel** thesis. Every interaction is supposed to make your understanding of a customer richer: the browse, the email open, the WhatsApp reply, the purchase, the return. That only compounds if those interactions get stitched to the same identity. A low match rate doesn't just cost you reach on one campaign, it breaks the flywheel at the source: data keeps accumulating, but it never consolidates into one sharper picture of the customer. You end up with more data and the same blurry understanding you had a year ago. And it connects to **early movers win channels**. Every platform shift of the last fifteen years, AdWords, Facebook, TikTok, WhatsApp, has rewarded whoever built solid foundations before the channel matured. As commerce keeps moving into messaging threads and [AI answer engines](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/), identity resolution becomes the prerequisite for showing up there at all. You can't personalize a WhatsApp journey or feed a recommendation into an AI shopping assistant for a customer you can't recognize. The brands solving match rate now are quietly building the foundation the next channel will run on. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260721-1436-why-your-match-rate-is-the-most-important-number-you-re-not--visual-2.png) ****Three vendor definitions of match rate** โ€” The same term, three different calculations, none of them apples-to-apples. ยท Data: artikel ## What to actually do about it: a short, practical checklist You don't need to buy anything to start. Match rate is measurable with what you already have. 1\. **Ask every ad platform you use what your current match rate is**, and how they calculate it. Most account managers can pull this; most marketers have simply never asked. 2\. **Get your definitions straight before comparing vendors.** Onboarding match rate, activation match rate, and graph match rate are related but not identical. Write down which one each vendor is quoting. 3\. **Audit your own identifier hygiene first.** Standardize email and phone formatting, deduplicate aggressively, and resolve known aliases (work email, personal email, loyalty ID) into a single profile before you blame the platform. 4\. **Run it as a holdout test, not a vendor claim.** If a vendor promises a match rate lift, split your audience, run the improved matching on one half, and measure incrementality against the untouched half. That's the same discipline I'd apply to any personalization claim, including my own. 5\. **Treat match rate as a standing KPI, not a one-time audit.** It decays as your customer base grows and identifiers splinter further. Check it quarterly, not once. None of this requires an AI model. It requires structured, unified, well-governed first-party data, which is exactly the boring, unglamorous prerequisite that thesis #2 keeps insisting on. ## Verdict Match rate is a real, underused metric, and Rokt mParticle deserves credit for putting a name on a problem most marketing teams feel but never measure. The structural pressure behind it (cookie deprecation, ATT, tightening walled gardens) is genuine and not going away. Where I'd hold back is on trusting any single vendor's percentage as gospel: the definitions aren't standardized, the lift claims aren't independently verified, and "improve your match rate" is a pitch every identity vendor in this space is making right now. My take: don't buy the fix before you've measured the problem. Audit your own match rate this quarter, on your own terms, with your own holdout test. If it's as low as the industry patterns suggest, you've just found a lever with more upside than another round of creative testing, and you'll know exactly what you're paying for when a vendor comes knocking :) Wrestling with identity resolution or match rate in your own stack? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Why the Conversation Is the New Funnel, Not the Feed URL: https://www.martechnext.com/why-the-conversation-is-the-new-funnel-not-the-feed/ Last updated: 2026-07-21T12:20:47.000Z **In this article:** - Why Spotify, YouTube, LinkedIn and Meta are all fighting the same "AI slop" fire, and losing on points - The uncomfortable math: platforms are financially incentivized to tolerate the flood - Why slop is a data flywheel problem, not just a content quality problem - Why messaging, chat and AI answers are becoming the cleaner alternative surface - A practical checklist for shifting budget and attention from feed to conversation - My verdict: don't abandon the feed, but stop betting your growth on it --- Picture a mid-sized DTC brand's social media manager opening LinkedIn on a Tuesday morning to check how a product launch post is performing. Between the notification and the actual post, she scrolls past four AI-generated "thought leadership" carousels, two obviously synthetic testimonial videos, and a meme account that has clearly never employed a human. She doesn't remember what she was looking for. Neither, probably, does the algorithm. That scene isn't fiction, it's just not attributed to anyone specific, because it's happening to millions of people a day, on every major platform, right now. Dutch outlet MT Sprout recently rounded up how the big platforms are responding: Spotify with detection and disclosure policies, LinkedIn with (unspecified, still-early) detection efforts, TikTok with creator education. The framing in that piece is right: AI slop is like weeds. It grows everywhere, and pulling it out by hand doesn't work when it regrows overnight. Here's the angle I want to add to that: this isn't just a content moderation story. It's a distribution story. And it's the strongest evidence yet for a thesis I've been making on MartechNext for a while: the conversation is becoming the new funnel, precisely because the feed is becoming unusable. ## The enforcement scoreboard, so far Let's look at what's actually been shipped, not just announced. Spotify says it has removed roughly 75 million "spammy" tracks and introduced new AI-disclosure and artist-protection policies (Spotify Newsroom, September 2025). YouTube tightened its Partner Program rules in July 2025 to explicitly restrict monetization of mass-produced, repetitive AI content. Ad-verification vendors DoubleVerify and Integral Ad Science have both launched dedicated products this past year: DoubleVerify's "AI Slop-Stopper for Social" and IAS's beta tool for blocking low-quality AI content near paid placements, building on their existing "Made for Advertising" detection tech. That's real movement. Three different types of players (a content platform, a video platform, and two ad-verification vendors) all independently concluded the problem was big enough to build dedicated tooling for. That alone tells you something: this isn't a moral panic, it's a resourcing decision made by companies that don't usually spend engineering time on problems that aren't costing them money. DoubleVerify's research also surfaced something worth remembering the next time someone tells you "AI will just get filtered out eventually": an operation researchers dubbed "AutoBait" used templated LLM prompts to mass-produce articles and images across more than 200 websites, purely to generate ad revenue. That's not a rogue creator experimenting with ChatGPT. That's an industrial process. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/draft-20260721-1335-van-spotify-tot-linkedin-techbedrijven-openen-de-jacht-op-ai-visual-1.png) ## The uncomfortable math nobody wants to fix Here's the part of the MT Sprout piece I think undersells the real story: none of this enforcement touches the actual incentive structure. TikTok's Creator Fund pays somewhere around $0.02 to $0.04 per 1,000 views across several major markets. That looks trivial per video. It stops looking trivial the moment you can generate hundreds of videos a day with zero marginal cost. Stanford researchers found that AI-image Facebook Pages, some running fully automated pipelines out of Pakistan, India, Vietnam, Thailand and Indonesia, generated hundreds of millions of interactions, with individual viral posts (the now-infamous "Shrimp Jesus" among them) reaching tens of millions of views. Advocacy group Fairplay estimated top AI-slop YouTube channels targeting children earned a combined $4.25 million in annual revenue (Fairplay, cited via Fortune, 2026), a figure YouTube itself disputes on methodology, and neither side has published an independently audited number. Meanwhile Meta's Creator Bonus Program has reportedly paid creators in markets including India and the Philippines for AI-generated content, even as Meta simultaneously develops its own generative ad-creative tools internally. That's not hypocrisy exactly, it's just what happens when the same company owns both the moderation lever and the growth lever, and growth is the one the CFO asks about. Let's be honest: engagement-based payouts and recommendation algorithms are the actual root cause here, and none of the announcements above touch them. YouTube's policy update is the closest thing to a structural fix in this whole list, and even that is a rule change, not an incentive change. This is Thesis 2 in practice, the right order, just applied at platform scale instead of a single company's stack. You can't clean up a mess with a detection layer bolted on top while the machine underneath keeps producing more mess faster than you can remove it. Structure first, then automate, then apply AI, applies to platforms fighting slop just as much as it applies to a retailer trying to personalize email. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/Screenshot-2026-07-21-at-14.14.15.png) ## Why this is a data flywheel problem, not just a quality problem Here's the piece that matters most for marketers specifically, and it's the one that gets the least attention in the coverage I've read. Recommendation and ranking algorithms learn from engagement signals. Slop generates engagement, sometimes more than genuine content, because it's optimized purely for the metric rather than for anything a human actually wants. Every time a slop post gets a click, a comment, a share, that signal feeds back into the model that decides what gets shown next. That's a negative flywheel. The data [flywheel](https://www.martechnext.com/from-rising-acquisition-costs-and-data-chaos-to-ai-driven-growth-in-travel/) thesis I keep coming back to on MartechNext usually describes something good: every interaction makes your system smarter, and that compounding advantage is the real moat, not the model. Slop inverts it. Every interaction with garbage content trains the ranking system on lower-signal, lower-trust data. The algorithm doesn't get smarter. It gets confidently wrong at scale. For a brand paying to appear in that feed, this isn't an abstract concern. If the platform's own signal quality is degrading, your targeting, your lookalike audiences, your attribution, all of it sits on top of a foundation that's quietly eroding. You can have a perfect creative and a perfect audience definition and still get worse results, because the environment measuring your performance is measuring itself against noise. **The feed's problem is becoming your measurement problem.** ## The conversation as the cleaner alternative This is where I think the MT Sprout framing, weeds that keep growing back, misses the more interesting strategic question: if the open feed is becoming structurally hard to keep clean, where does commerce and discovery go instead? My answer, and the position MartechNext has staked out for a while now: into messaging, chat, and AI answer surfaces. WhatsApp threads. [Meta's Business Agents](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/). ChatGPT product answers and the emerging discipline of [generative engine optimization](https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/). These aren't immune to spam by default, but they're structurally harder to flood, because they're one-to-one or small-group, permissioned, and identity-anchored rather than open and algorithmically amplified. A slop farm can post a thousand AI images to a public feed for free. It's much harder and much less profitable to spam a thousand individual WhatsApp threads where the recipient has to have opted in first. Consider a simple, round-number comparison. Say a travel brand spends its media budget the traditional way: a โ‚ฌ10,000 monthly budget on paid social, competing for attention against an increasingly slop-saturated feed where organic reach keeps declining and CPMs keep climbing because platforms need to backfill declining engagement quality with more ad inventory. Compare that to the same โ‚ฌ10,000 invested in building out a WhatsApp journey (from booking confirmation to pre-trip tips to post-trip re-engagement) where every message lands in a channel the customer actively chose, at a 90%+ open rate territory that's genuinely differentiated from feed reach, even if we should always be careful, as I've flagged before, that channel statistics like open rates aren't conversion statistics on their own. Full disclosure, as always: I'm the founder of Zoey, a WhatsApp journey automation platform for the travel industry, so I have skin in this exact game. That's precisely why I want to be honest about the caveat above rather than let the open rate number do all the work. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/Screenshot-2026-07-21-at-14.13.57.png) Feed versus Conversation ## What to actually do about it You don't need to abandon feeds. Most marketers still need reach, and reach still mostly lives in feeds for now. But the balance of investment needs to shift, and there are concrete steps you can take this quarter: 1\. **Audit your adjacency risk.** If you're running programmatic or social spend at any scale, look at DoubleVerify's or IAS's AI-content adjacency tools before you assume your ads aren't sitting next to AutoBait-style slop farms. 2\. **Stop trusting platform-reported quality metrics at face value.** Spotify's 75 million figure, YouTube's enforcement claims, they're all self-reported, with no independent audit of false positive or false negative rates. Treat them as directional, not gospel, the same way I'd want you to treat my own case studies. 3\. **Move discovery budget toward owned, permissioned conversation channels.** WhatsApp, email with genuine two-way interaction, in-app chat. Not because feeds are dying tomorrow, but because the ROI curve on feed reach is bending the wrong way while conversation channels are still underpriced. 4\. **Treat GEO and feed-quality auditing as the same discipline.** Whether it's making sure your product gets surfaced correctly in a ChatGPT answer or making sure your ad isn't parked next to synthetic garbage, it's the same underlying job: verifying the environment your brand appears in, because you can no longer assume the platform is doing that verification reliably on your behalf. 5\. **Build for one-to-one, not for algorithmic amplification.** Recommendation and ranking systems reward volume and engagement, which is exactly the incentive slop exploits. [A well-built conversational journey rewards relevance to one person](https://www.martechnext.com/the-power-of-intelligent-personalization-how-ai-will-transform-the-customer-journey/), which is a much harder thing to fake at scale. ## My verdict Spotify, YouTube, DoubleVerify and IAS deserve real credit here. They're building genuine detection capability faster than most people expected, and Thesis 7 applies: whoever gets AI-slop verification right first, in ad tech specifically, is going to own a category that's about to become mandatory infrastructure rather than a nice-to-have. But detection capability is not the same as control, and control is what determines whether marketers can actually trust these environments again. That's Thesis 5 in its purest form: the capability to detect slop is racing ahead, the willingness (and financial incentive) to actually remove it at the root is lagging badly, and until platforms touch the engagement-based payout structures underneath, this is going to feel like mowing weeds that grow back overnight, exactly as the original MT Sprout piece put it. My money says the open feed doesn't get fixed in the next two years. It gets partially cleaned up on the surface while the underlying incentive rot continues, and the smart marketers respond not by waiting for platforms to solve it, but by quietly building real strength in the channels slop can't easily reach. Wrestling with where to shift budget between feed and conversation in your own stack? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### CM.com's profit jump is proof the conversation is becoming the funnel, not just a feature URL: https://www.martechnext.com/cm-coms-profit-jump-is-proof-the-conversation-is-becoming-the-funnel-not-just-a-feature/ Last updated: 2026-07-21T08:44:30.000Z Most of what I write about in this newsletter is funding rounds: promises, theses, money committed to a bet that hasn't paid off yet. This one is different. CM.com [just posted numbers](https://www.cm.com/cdn/web/en/file/investor-relations/cmcom-q2-2026-press-release.pdf?ref=martechnext.com), real ones, and they happen to validate something I've been arguing for a while: the conversation is becoming the funnel, not a feature bolted onto it. That matters because conversational commerce has lived mostly on vibes so far. Vendors love to talk about WhatsApp open rates and chatbot engagement. What we rarely get is a company at real scale showing that the model actually makes money. CM.com just did. ## The details CM.com, the Breda-based communications and martech provider behind the Halo platform, reported second-quarter results this week. Revenue grew 14 percent year over year to 67 million euros, up from 9 percent growth in Q1, so the growth rate is accelerating, not just holding steady. Adjusted EBITDA rose 61 percent to 6.3 million euros. Gross profit came in at 20.5 million euros, up 3 percent, recovering from a small dip earlier in the year. There's still a net loss of 1.1 million euros on the bottom line. But compare that to the 4.6 million euro loss in the same quarter last year, and the trend line is unmistakable. CEO Jeroen van Glabbeek called it "clearly improved operational leverage." The company maintains its full-year guidance of over 30 percent adjusted EBITDA growth. ## The MartechNext take Let's apply the skepticism checklist for a second, because I do that to everyone, including companies I'm rooting for. Is this a per-metric flattering itself? Not really. Revenue, EBITDA, and gross profit are top-line, verifiable numbers, not "8x engagement" style vendor math. Is there a selection effect? Also not much of one here: this is a public company reporting quarterly results, not a case study cherry-picked from a happy customer. Is the "30 percent EBITDA growth" guidance vendor spin? It's forward guidance, so treat it as a target, not a fact, but the direction of travel across three straight quarters is hard to argue with. So what does this actually tell us? CM.com's business is built on messaging infrastructure: SMS, WhatsApp Business API, voice, and increasingly the Halo platform, which bundles conversational AI with commerce and service workflows. When a company like this shows accelerating revenue growth *and* improving margins at the same time, that's the tell. It's not just growing by throwing more sales reps at more logos (that shows up as revenue growth with flat or worsening margins). Operating leverage improving alongside revenue acceleration usually means the infrastructure is doing more of the work per euro spent. That's what maturing platform economics look like. This is the same thesis I've written about with [Meta's Business Agents ](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/)and in the [WhatsApp Marketing 2026](https://www.martechnext.com/whatsapp-marketing-in-2026-what-marketers-need-to-know-to-use-it-effectively/) piece: the thread is the new funnel. Commerce is moving into messaging, and the companies that provide the plumbing for that shift are the ones to watch, more than the AI chat wrapper of the month. CM.com isn't glamorous. It doesn't have a slick AI narrative or a nine-figure Series B behind it. It's a communications company that's been building conversational infrastructure for over two decades and is now seeing the market catch up to the thesis. Here's the honest caveat: one company's quarter is not an industry trend. CM.com operates across many verticals and geographies, and its Halo platform is one of several bets in a broader portfolio that includes SMS and payments. I'd want to see this pattern repeat for a few more quarters, and ideally see it echoed by other conversational commerce players (Sinch, Infobip, and the WhatsApp BSP ecosystem broadly) before calling it a category-wide inflection point. A single earnings beat is a data point, not a trend line. But it's a genuinely useful data point, because most of what we get in this space is vendor-reported case studies with selection bias baked in (the same critique I'd apply to my own published cases, for the record). Public company financials are audited, comparable quarter over quarter, and impossible to cherry-pick. When one of the few publicly listed conversational commerce infrastructure providers shows accelerating growth and improving margins in the same breath, that's worth more than a dozen glossy funding announcements. **Verdict:** small quarter, meaningful signal. If you're building or buying into conversational commerce infrastructure, this is the closest thing to hard proof yet that the category is past the hype phase and into the "actually makes money" phase. Keep an eye on Q3 :) ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/visual-1-1.svg) \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Your ad platform now takes orders from someone else's AI agent. Are you ready for that? URL: https://www.martechnext.com/your-ad-platform-now-takes-orders-from-someone-elses-ai-agent-are-you-ready-for-that/ Last updated: 2026-07-21T08:35:08.000Z **In this article:** - What adpaq actually built, and why an MCP server is different from a normal API - Why this is a bigger deal than the size of the announcement suggests - The real question: who is accountable when an agent makes the wrong call - How this connects to the trust/control curve that decides deployment speed - What to check before you let any agent touch your ad accounts - A verdict on whether this is signal or noise Picture a media buyer at a mid-sized agency, typing a brief into a chat window instead of opening a campaign dashboard: "Set up a 15-second audio spot for the Amsterdam market, target commuters aged 25-45, โ‚ฌ8,000 budget, start Monday." No login, no line-item form, no dropdown menus. The agent takes it from there: builds the campaign, sets targeting, links creatives, and waits for a human to click approve on the budget. That's not a demo reel. It's what a Dutch programmatic audio DSP just shipped. ## The details adpaq, a Dutch demand-side platform for programmatic digital audio, has built its own Model Context Protocol (MCP) server. MCP is the emerging open standard that lets AI agents not just retrieve information from a system, but take controlled actions inside it. adpaq says this makes it the first platform in programmatic audio to let AI agents work directly inside the DSP, briefed in natural language. Concretely: an agent can prepare a full audio campaign, create line items, set targeting by location, device, publisher, or audience, attach creatives, and generate performance reports. The rollout is a partnership, not a wide release. adpaq is running the first implementation with Abovomaxlead, which it describes as the largest independent media and marketing agency in the Netherlands (source: Emerce, July 2026). The two are testing the technology under real campaign conditions before adpaq opens it up to its broader partner network. adpaq has built in a permission layer: informational tasks, like pulling reports or checking pacing, can be fully agent-driven. Sensitive actions, budget changes or campaign deletion, always require explicit human approval. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/visual-1.svg) ## Why an MCP server is not just a fancier API If you've built or bought martech integrations before, your first instinct might be: didn't we already have APIs for this? Fair question. Here's the actual difference. A traditional API is built for developers to write fixed, predictable integrations. You know exactly what call does what, ahead of time. An MCP server is built for an AI agent to discover what's possible on the fly, in natural language, and decide which actions to chain together to reach a goal it wasn't explicitly told the steps for. That distinction matters because it's the difference between generative AI and agentic AI, and adpaq's own framing gets this right: generative AI mostly helps you write or analyze; agentic AI acts, autonomously, across multiple steps, to reach an outcome. Setting up a campaign isn't one action. It's targeting, creative linking, budget allocation, and reporting, chained together. That chaining is exactly what MCP is designed to let an agent do safely. This is why MCP adoption in martech is worth watching closely, even when the individual announcements look small. It's infrastructure, not a feature. ## The MartechNext take Let's connect this to a thesis I keep coming back to: **automation with human accountability**. AI should handle the operational layer, the repetitive campaign mechanics, while humans keep the judgment calls, the taste, the relationships. What decides how fast any of this actually gets deployed isn't how capable the AI is. It's how much control the humans around it are willing to give up. I've called this the trust/control curve lagging the capability curve, and adpaq's permission split is a textbook illustration of it. Look at what they actually gated. Reporting, targeting mechanics, creative linking: agent territory. Budget changes and deletions: human-only. That's not arbitrary. It's the two categories of action that, if an agent gets them wrong, cost you money you can't easily claw back versus mistakes you can catch and fix in the next report. That's a sensible first cut at the control problem, and it's more thoughtful than a lot of "AI-powered" platform launches I've seen that bolt a chatbot onto a dashboard and call it agentic. **What's genuinely strong here:** adpaq didn't wait for a big platform to define the standard for them and then bolt on a wrapper. They built their own MCP server, ahead of a standard still very much in motion, in a niche (programmatic audio) where AI-native tooling isn't crowded yet. That's the early-mover instinct that has paid off in every previous platform shift, AdWords, Facebook Ads, TikTok, WhatsApp Business API. The companies that build the foundation before the channel matures are the ones still standing when it does. A Dutch DSP getting there before the bigger ad tech players is a genuinely interesting bet. **Now let's apply some scrutiny, because that's the job.** First: "first in programmatic audio" is a category claim, and category claims are almost always narrower than they sound. First in programmatic *audio* is a much smaller claim than first in programmatic advertising, and audio is a relatively small slice of the ad tech market to begin with. That doesn't make it worthless, being early in a niche is exactly how you build a moat before anyone bigger notices, but don't read "first" as "leading the industry." Second: there are zero performance numbers in this announcement, and that's actually refreshing, but also a reminder to wait. No claim yet about how much faster campaign setup gets, no claim about how agent-built campaigns perform against human-built ones. Good, because if adpaq had shipped a number like "campaigns launch 60% faster" on day one of a partner pilot, I'd apply the same checklist I apply to every vendor stat: what's the baseline, who's the comparison group, and is the pilot agency's team more AI-fluent than the average buyer (a classic selection effect)? The absence of numbers here isn't a red flag. It's honesty about being in a validation phase. Watch what numbers show up once the pilot concludes, and ask those questions then. Third: "the user always stays in control" is the kind of line every agentic AI vendor says right now, and it's worth being precise about what it actually means in this case. Here it means something specific and checkable: budget changes and deletions require approval. It doesn't mean a human reviews every targeting decision or every creative pairing the agent makes. That's a meaningfully narrower version of "control" than the phrase implies at first read, and it's worth asking any vendor making this claim to specify exactly which actions are gated, not just to accept the reassurance. Fourth, and this is the one agency buyers should sit with: if an agent can build a full campaign from a one-line brief, what exactly is the campaign specialist doing next year that they're doing today? adpaq's own answer, working out how AI can "collaborate optimally" with campaign specialists (their words, paraphrased), is the right question to be asking. But it's an open question, not a solved one. The honest answer right now is: nobody fully knows yet, including adpaq. That uncertainty is normal at this stage of a technology, not a knock against the announcement. ## What to check before an agent touches your ad accounts If your agency or in-house team is evaluating agentic AI tools for ad platforms, whether it's adpaq's MCP server or a competitor's equivalent, here's a short list worth running through before you plug anything in: 1\. **What's gated, exactly?** Get the specific list of actions that require human approval versus actions the agent can execute unsupervised. Don't accept "you're always in control" as an answer on its own. 2\. **What's the audit trail?** If an agent sets a wrong targeting parameter, can you see exactly what it did and why, after the fact? 3\. **What happens on ambiguous briefs?** A one-line brief has gaps. Does the agent ask clarifying questions, or does it guess and proceed? 4\. **Who's accountable for a bad outcome?** If an agent-built campaign underperforms or overspends inside its approved budget, is that a platform issue, an agency issue, or a "well, you approved the budget" issue? Get this in writing before, not after. 5\. **Is this reversible fast?** How quickly can a human pause or roll back an agent action once it's live? None of these are reasons to avoid the technology. They're the due diligence that separates "we adopted agentic AI" from "we adopted agentic AI and it went fine." ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/Screenshot-2026-07-21-at-10.32.53.png) ## The bigger pattern: MCP is becoming ad tech plumbing This isn't an isolated story. Model Context Protocol adoption is moving through martech the way REST APIs did a decade ago, quietly, platform by platform, until one day every serious tool has one and you can't remember a time they didn't. adpaq is early. It won't be the last DSP, ESP, or CDP to ship one this year. That connects directly to a thesis I've written about before: the conversation is becoming the interface layer for commerce and, increasingly, for the tools marketers use to run commerce. [(See: Meta Business Agents 2026) ](https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/)We've mostly talked about this in the context of customers chatting with brands. adpaq's announcement is the same shift happening one layer up the stack: marketers chatting with their own platforms instead of clicking through them. ## Verdict Small, narrow rollout. Genuinely early signal. adpaq isn't claiming to have solved agentic AI in ad tech, and the fact that they're running a controlled pilot with one implementation partner before opening it up is the right level of caution for infrastructure this new. The permission model (informational tasks to the agent, financial and destructive actions to the human) is the correct starting split, and it's a useful reference point for anyone building or buying similar tooling. But the honest read is: this is a bet on where the industry is heading, not proof that it's arrived. The real test is what happens during the Abovomaxlead pilot, whether agent-built campaigns hold up on reliability and usability, not just speed. My take: this is worth watching, not worth over-hyping. The companies that get MCP integrations right, with sensible guardrails, before the standard fully matures, are the ones who'll own the workflow when every buyer expects to brief a campaign the way they'd brief a colleague. adpaq just made an early, sensible move in that direction. Whether it pays off depends entirely on execution over the next few quarters, not on the announcement itself. Watching how your platforms handle agentic AI, or wrestling with where to draw the human-approval line in your own stack? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Invisible to travelers in ChatGPT? How to make sure your travel brand gets found URL: https://www.martechnext.com/invisible-to-travelers-in-chatgpt-how-to-make-sure-your-travel-brand-gets-found/ Last updated: 2026-07-20T09:04:38.000Z **In this article:** - Why AI platforms are closing in on search engines as the starting point for travel research - GEO (generative engine optimization) explained, and why it's SEO in new clothes rather than a new discipline - The biggest risk almost nobody manages: inconsistent and outdated information - Co-occurrence: the new backlink, and the new role for destination organizations - Why direct suppliers currently beat the OTAs in AI answers, and why that window is closing - A practical roadmap for teams with limited capacity It usually starts with a small moment inside a travel company. An employee types the name of their own hotel or travel brand into ChatGPT, asks for a recommendation for the destination, and watches the competitor get named while their own brand doesn't appear. The question that follows is always the same: how do we fix this? That's the exact moment AI visibility stops being a vague future theme and becomes an acute problem. And travel is the industry that needs to get to work on it now. ## Travelers research trips through AI now Let's start with the numbers, because they leave little room for doubt. According to Phocuswright, usage of generative AI platforms like ChatGPT and Gemini for travel research has grown fivefold in a single year. They now sit at 33%, just behind classic search engines at 35%. Measured more broadly, 56% of travelers used AI for at least one trip in the past twelve months, up from 33% in early 2025\. Phocuswright calls it the fastest behavioral shift they've measured in over a decade of consumer research, with every generation posting double-digit gains. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/geo-ai-vs-search.png) AI is catching up with the search engine Adoption is not evenly distributed, and that's useful information rather than a footnote. The US leads. In Europe, adoption trails partly due to stricter regulation slowing feature rollouts, with the UK ahead of France and Germany. But everywhere the curve points the same way. If you operate in a market where the numbers still look modest, read the US data as a preview, not as someone else's problem. Waiting until adoption in your market matches the US means entering the game when everyone else is already on the field. A scene from a recent industry conference illustrates the gap perfectly. Speakers asked a room full of travel professionals who had a strategy for AI visibility. Two hands went up. Then they asked who had personally researched a trip with an AI tool. Every hand in the room went up. That's the opportunity in one image: the usage is mainstream, the strategy is not. And there's good news buried in the data. Visitors who land on a travel website via ChatGPT or Claude have largely done their research already. They arrive better qualified, and it shows in conversion. Practitioners I've spoken with saw AI-referred traffic grow five to eight times in a year. Still from a small base, but the growth curve is unmistakable. Half of travelers who see AI answers still click through to the source websites, and that click is worth more than the average search click ever was. ## GEO is not a new SEO. It's SEO in new clothes Here's where the first mistake gets made. Plenty of vendors present AI visibility as a completely new discipline, disconnected from everything you already know. The reality is more nuanced. AI visibility, usually called GEO (generative engine optimization), still leans heavily on the principles of SEO. The difference is in the execution, not the foundation. Google itself, tellingly, still just calls all of it SEO. What makes it more complex is that every AI model works slightly differently. ChatGPT leans heavily on the Bing index for its web search, while Gemini correlates much more closely with Google's rankings. A strong position in Google doesn't automatically translate into visibility across every AI platform. And everything moves fast: what works today can shift within months. We're roughly where SEO stood fifteen to twenty years ago. Nobody knows exactly what drives visibility, and the moment you think you've figured it out, it changes. That doesn't mean you can't act. It means you work from proven principles and monitor relentlessly. Anyone declaring SEO dead has it backwards: SEO is the foundation GEO stands on. ## Write for the machine and the human The biggest practical shift is in how you structure content. Think of the classic online pancake recipe: before you reach the ingredients, you scroll through a long story about the writer's grandmother. For a human, charming. For an AI model, noise. Many travel websites still work exactly like this: long, atmospheric landing pages with the relevant information buried somewhere in the middle. AI models work within a limited window of information. They prefer short, clear blocks of content in which a question gets answered directly. In practice this is called content chunking: you break your content into compact, self-contained blocks that an AI can easily pick up and cite. At the same time, thinking shifts from individual keywords to entities. You no longer just want to rank for "boutique hotel Amsterdam"; you want the AI model to associate your brand with the right concepts, atmospheres, and destinations. The foundations that support this are partly familiar from SEO: a fast, well-indexed website, clean structured data so models can identify your brand, products, and location, a logical heading structure, and FAQ blocks that answer concrete questions. On top of that, strong E-E-A-T signals (Experience, Expertise, Authoritativeness, Trustworthiness) weigh heavily, because AI systems prefer sources they recognize as reliable. ## Your biggest risk: contradictory and outdated information This might be the most important point in this whole piece, because it's so concrete. AI models are extremely literal. Where a human effortlessly ignores an outdated page, an AI takes it at face value. Still selling your Easter promotion in May? Does your site say the pool opens in October when it's already open? Is your property name consistent everywhere? These contradictions are a far bigger risk than most companies realize. A real example: a hotel renamed one of its properties, updated the website neatly, but forgot the metadata. Overnight, the hotel vanished completely from AI answers, simply because the name the model was looking for no longer matched what the website reported. Only when the company noticed it had disappeared from the answers did anyone ask: what happened to our hotel? ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/geo-inconsistency.png) One inconsistency, and you vanish from AI answers In travel this happens more often than you'd think. Properties change brands and chains all the time. For SEO, everyone knows how to handle that with redirects. For GEO, almost nobody thinks about it yet. The good news: you can absolutely communicate such a change properly toward the AI models. You just have to know it's necessary. So always start with a check on the reliability and consistency of your own information, before you invest a euro in visibility. ## Co-occurrence: the new backlink In the old SEO world, backlinks were the core measure of authority. In the world of AI visibility, something similar is emerging: co-occurrence. In short, it's valuable when authoritative sources mention your brand in the same context as a topic. If your travel company is named in a piece about the best treks in a region, that helps your visibility, even without a classic link. This creates a genuinely new role for destination marketing organizations and tourism boards. A DMO no longer needs to steer purely on visitor numbers or clicks to its own site. Its biggest lever may be actively naming the businesses in its region and functioning as a source. That creates a cycle: the organization strengthens its region, and the affiliated travel businesses get found. For an independent supplier, the practical implication is clear: actively seek partnerships with parties that can mention you in the right context. And in this phase of the journey, don't forget your Google Business Profile. A well-optimized profile demonstrably helps visibility at the moment the traveler makes a concrete choice. ## Direct versus OTA: your window of opportunity Now the strategically most interesting part. Who wins in this new landscape, the supplier or the big booking platforms? Right now, direct travel suppliers hold a remarkable advantage. AI models frequently recommend the supplier itself rather than the OTAs. The reason is logical: the content an AI needs is richer and more specific when it comes from the supplier. An OTA optimizes for simplicity and comparability, and that's often not what convinces an AI model. It matches how travelers search now. The query is no longer "hotel Barcelona center" but something like "I'm looking for a boutique hotel in a lively neighborhood, close to good seafood restaurants". That kind of nuanced, context-rich information rarely lives neatly structured in an OTA database, but it does live in the supplier's own descriptions. There is a timer on this, though. Industry experts estimate that direct suppliers have roughly eighteen months before the big players, Booking and Expedia first among them, close this gap. They have enormous, capable teams working on exactly this, while hedging by shifting toward B2B and infrastructure roles. The message is clear: the advantage is real, but temporary. Whoever gets their AI visibility in order now wins back customer relationships that will be much harder to capture later. For why owning that direct relationship matters so much economically, see my piece on the shift from acquisition costs to AI-driven growth in travel (link it here). ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/geo-window-of-opportunity.png) You window of opportunity ## So is the website obsolete? A fair question, because you increasingly hear that websites will disappear. The opposite is true. Everything above revolves around your website. It has simply gained a second audience: alongside your guest, the AI models now read along. Your website is where you communicate with those models and where your information comes from. That makes it more important, not less. The data backs this up. Only 8% of travelers found an AI answer sufficient on its own, and half click through to the source website after an AI answer. Travel is a high-stakes purchase where people want to verify. Behavior also shifts slower than headlines suggest: eighteen months ago we all assumed we'd soon be booking entirely in chat, and in practice that turned out to be an unpleasant experience for a complex trip. The traveler may now do 80 to 90% of the dreaming and orienting through AI, but still takes the final, decisive step on a website built for conversion. And even if the interface eventually shifts to voice or a wearable, there will always need to be a source the AI pulls its information from. ## Where to start: a practical roadmap The biggest hurdle is rarely willingness, it's capacity. In many travel companies the same person is responsible for SEO, GEO, CRM, and often paid media on top. So this isn't about doing everything at once, but about the right order and the highest-impact steps. Three starting points: **1\. Start with monitoring.** You don't need an expensive tool on day one. Open ChatGPT yourself and ask the questions your guest would ask. See what comes out, and whether you're in it. Mind one detail most people miss: over 80% of users are on the free model, and the answers of the free and paid models differ substantially. With web search turned off, you get answers based on outdated training data. For a serious picture, monitor deliberately and consistently, with the right models and settings. **2\. Get the technical basics in order.** Check that you're not unintentionally blocking AI crawlers, ensure correct structured data, a logical site structure, and accurate, consistent information. It's the least sexy part of the work, and it's what prevents your investments from leaking away. **3\. Work from a prioritized roadmap.** There are hundreds of small things you could do. The craft is determining which ones have the most impact today: the low-lift, high-impact optimizations. And don't wait for perfect data. The dream of one clean, complete customer database has existed for twenty years and remains unrealistic for most travel companies in the short term. It's far smarter to work within your current data limitations and focus on what actually matters to the customer. ## Ready to be found in AI? The through-line is clear: travelers are already in the AI platforms en masse, most travel companies are not, and there is still a window in which you can build a lead. The companies that get their foundation in order now, make their information consistent, and optimize deliberately for both Google and the AI models are the companies that will be named. This isn't a separate trick next to your marketing, it's the same discipline that has always mattered: being findable where your customer searches. The place where they search has simply moved. :) As always, I don't have all the answers, and in a field this young, anyone who claims to should make you suspicious. But the principles above are the ones I see working in practice today. Wrestling with AI visibility in your own organization, whether in travel or elsewhere? I'd love to exchange thoughts. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Meta Business Agents: your customers will soon buy in the chat, and they may never see your website URL: https://www.martechnext.com/meta-business-agents-your-customers-will-soon-buy-in-the-chat-and-they-may-never-see-your-website/ Last updated: 2026-07-20T07:31:37.000Z **In this article:** - What Meta Business Agents are, and why they're a step beyond the classic chatbot - The two tiers: the free self-serve agent versus the enterprise Business Agent Platform - WhatsApp Business Discovery: the search engine hiding inside a messaging app - Why this is part of a bigger shift, from OpenAI's product feed ads to the pressure on the product page - A practical to-do list for marketers, wherever you operate Imagine your best customer next quarter: she discovers your brand through the WhatsApp search bar, asks a few questions in the chat, browses your product feed, and completes the purchase. All in one thread. She never opens your website, never sees your carefully optimized landing page, and never touches your checkout flow. That scenario stopped being hypothetical at Meta Conversations 2026 in London, where Meta introduced the Business Agent: autonomous AI agents that guide customers through support, product discovery, and checkout inside a single WhatsApp, Messenger, or Instagram thread. I covered the announcement itself in my earlier [analysis of Meta Conversations 2026](https://www.martechnext.com/meta-conversations-2026-the-business-agent-is-here-and-it-changes-your-whatsapp-strategy/). This piece zooms out: what has happened since, what it means internationally, and what you should actually do about it. ## From chatbot to agent: what changed The difference between a Business Agent and the chatbots you know is fundamental. Rule-based bots follow scripts and break the moment a customer goes off-script. Business Agents understand context, hold multi-turn conversations, respond in the customer's language while keeping your brand voice, and, crucially, take action: checking real-time inventory through API connections, qualifying leads, booking appointments, and walking a customer through payment. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/introducing-meta-business-agent.png) The agent learns from your own materials: chat history, website, product catalog, and Facebook page. Meta says more than one million businesses were already using a version of the agent before the global announcement, on top of more than one billion active business threads per day across WhatsApp, Messenger, and Instagram. That installed base is exactly why this launch matters more than any startup demo: the customers are already in the channel. The agent just removes every reason to leave it. ## Two tiers, and a meter that starts running In practice, Meta launched two different products under one name, and the distinction matters for your planning. **The self-serve tier** lives directly inside the WhatsApp Business app: set up in minutes, train it on your existing content, and it responds to inbound conversations around the clock. It's deliberately simple, which also means limited. It handles inbound only, doesn't send broadcast campaigns, and doesn't connect to your CRM. Think of it as the entry drug for automated messaging: millions of small businesses will get their first taste of conversational AI here. **The Business Agent Platform** is the enterprise version, currently in limited early access. This is where it gets interesting for serious commerce: connections to hundreds of systems like Shopify and Zendesk, enterprise-grade guardrails, custom rules, and measurement built in. This tier is what turns the agent from a support tool into an autonomous commerce layer. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/business-agent-works-for-you-too.png) And note the pricing timeline. The free testing window that opened on July 1 closes soon: from August 1, 2026, Business Agent conversations are billed per token, at $2.00 per million tokens, roughly 4 to 5 cents per conversation. A second pricing wave follows on October 1\. Two takeaways: first, if you want to experiment cheaply, the window is closing. Second, the unit economics are remarkable. At a few cents per handled conversation, the cost comparison with human support teams, or even with outsourced chat agencies, isn't close. ## The quiet headline: WhatsApp becomes a search engine The feature that received less attention than it deserves is WhatsApp Business Discovery: users can now find businesses directly through the WhatsApp search bar. Combine that with the new sharing mechanics (someone shares your business in a chat, the recipient starts a conversation with one tap; someone shares a restaurant, directions are available inside the same thread) and you're looking at something we should name for what it is: a new search engine with unusually high purchase intent, inside one of the most used apps on the planet. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/discover-ai-powered-businesses-1.png) For markets where WhatsApp dominates daily communication (Brazil, India, Indonesia, Mexico, large parts of Europe, Africa, and the Middle East), this is not a niche feature. Being findable in WhatsApp search will become a discipline in its own right, the way being findable in Google became one twenty years ago. Early movers will enjoy the same advantage they always do: less competition for the same intent. ## The bigger pattern: the purchase is moving into the conversation Meta is not acting in isolation. Look at what happened in the past few months across the industry: OpenAI opened up product feed ads in ChatGPT, letting retailers connect their product catalogs so sponsored products appear directly inside conversations, targeted on conversational intent rather than search history or browsing behavior. The setup deliberately mirrors Google Shopping, down to the feed schema, so merchants can reuse the structured product data they already maintain. Agentic checkout support through platforms like Shopify closes the loop from question to purchase. Add Google's own moves toward agent-led shopping experiences, and the direction is unmistakable: every major platform is racing to keep the entire journey, from discovery to payment, inside its own conversational surface. The classic product page, the destination we've spent two decades optimizing, is under structural pressure. Not dead, but no longer the only place where buying happens. For marketers, this creates a new discipline that sits somewhere between SEO, feed management, and CRM. Call it conversational commerce optimization: making sure your products, answers, and brand voice are machine-readable and agent-ready, wherever the conversation happens. ## What you should do now Wherever you operate, the homework is the same. Four priorities: **1\. Get your product feed in order.** Your feed is no longer just fuel for Google Shopping and Meta catalog ads. It's what a Business Agent serves to customers inside the chat, and what ChatGPT reasons over when someone asks for "running shoes under $100". Complete titles, accurate availability, rich descriptions: feed quality is now customer experience, not backend hygiene. **2\. Treat WhatsApp discoverability like SEO.** Complete your business profile, think about the terms customers would search for, and monitor what shows up. This field barely exists yet, which is precisely why it's worth your attention now. **3\. Guard your brand voice in automated conversations.** An agent that speaks for your brand around the clock, in multiple languages, is a wonderful thing right up until it isn't. Set guardrails, define escalation rules, and test relentlessly. As I wrote in my analysis of the announcement: autonomy without control is not a strategy, it's a gamble. **4\. Fix your measurement before the revenue moves.** Conversions that happen inside WhatsApp or Instagram threads don't show up in your web analytics. If a growing share of your journey completes off-site, your reporting will quietly understate exactly the channel that's growing fastest. Decide now how you'll attribute conversational revenue. For a broader foundation on the channel itself (formats, GDPR and consent rules, and how to build your first WhatsApp strategy), see my [complete overview of WhatsApp marketing in 2026](https://www.martechnext.com/whatsapp-marketing-in-2026-what-marketers-need-to-know-to-use-it-effectively/). ## Conclusion: the thread is the new funnel The through-line of Meta Conversations 2026 was the commercialization of messaging, and everything since has confirmed it. Meta is shifting the emphasis from businesses reaching out to potential customers toward consumers discovering and buying from businesses directly in the chat. WhatsApp stops being a service channel bolted onto your funnel; increasingly, it is the funnel. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/business-agent-responds-to-customers-24-7.png) The history of digital channels keeps repeating itself: the companies that build their foundation before the channel matures get years of cheaper attention and less competition. The Business Agent's free window is closing, Discovery is barely contested, and most of your competitors are still treating messaging as a support cost center. That combination won't last. :) Full disclosure, as always: I'm the founder of Zoey, a WhatsApp journey automation platform for the travel industry, so I have skin in this game. It's also exactly why I follow this space as closely as I do. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Kimi K3: the first open 3-trillion-class model, and why marketers should pay attention URL: https://www.martechnext.com/kimi-k3-the-first-open-3-trillion-class-model-and-why-marketers-should-pay-attention/ Last updated: 2026-07-17T10:26:28.000Z Moonshot AI released Kimi K3 today, and it's a milestone for open models: 2.8 trillion parameters, making it the first open model in the 3T class. It comes with native vision, a 1-million-token context window, and benchmark scores that put it right among the frontier. The full model weights will be released by July 27. ## The details Kimi K3 is built on two architectural updates, Kimi Delta Attention and Attention Residuals, combined with an aggressively sparse Mixture of Experts setup that activates 16 of 896 experts. Moonshot claims roughly 2.5x better scaling efficiency than its predecessor Kimi K2, meaning the model converts compute into capability more effectively. On coding benchmarks, K3 trades blows with the best proprietary models. It tops Program Bench and SWE Marathon, sits within half a point of GPT-5.6 Sol on Terminal Bench 2.1, and lands between Claude Fable 5 and the rest of the field on FrontierSWE. Moonshot is refreshingly candid about the overall picture: by its own admission, K3 still trails Claude Fable 5 and GPT-5.6 Sol in general performance and user experience. But for an open model, this is uncharted territory. Beyond coding, the launch leans heavily on agentic knowledge work. Moonshot showcases K3 producing consulting-style industry reports with interactive visualizations, publication-ready slide decks, spreadsheet automation, and even video editing: the model cut its own teaser video from 56 source clips, work that would take an experienced human editor a day or two. Pricing through the Kimi API: $0.30 per million cache-hit input tokens, $3.00 cache-miss, and $15.00 per million output tokens. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/kimi-k3-coding-benchmarks.png) ## The MartechNext take You might wonder why a Chinese frontier model release belongs on a martech platform. Three reasons. **The price of intelligence keeps collapsing, and that resets the martech value equation.** Every time an open model closes the gap with the proprietary frontier, the model layer becomes more of a commodity. For marketing teams and martech vendors, the strategic implication is the one I keep repeating: the model is not your moat. Your data foundation, your customer profiles, and your workflows are. If frontier-adjacent intelligence costs $3 per million input tokens and the weights are downloadable, the differentiation question shifts entirely to what you feed it and how you orchestrate it. **Open weights matter for European marketers specifically.** A model you can self-host is a model you can run inside your own infrastructure, with your customer data never leaving your environment. For personalization use cases that touch first-party data, that's a meaningful compliance and privacy story. I expect a wave of martech vendors quietly swapping in open models for exactly this reason, and "which model runs under the hood" becoming a procurement question rather than a marketing bullet. **The knowledge-work demos are the agency-replacement story again.** Look past the coding benchmarks and you see what Moonshot is really selling: an agent that produces research reports, slide decks, dashboards, and edited video autonomously. That's the same structural shift I wrote about with Birdsview in retention marketing, now aimed at the content and insights layer. The work martech agencies bill for is increasingly the work these systems demo for free. Now the caveats, because launch-day benchmarks deserve them. These numbers are vendor-run, with different agentic harnesses per model, and the comparison footnotes read like a legal document. Directionally credible, but not gospel. More interesting to me is what Moonshot lists under limitations: K3 shows "excessive proactiveness", meaning it can make unexpected decisions on the user's behalf when instructions are ambiguous. For a coding agent, that's an inconvenience. For an autonomous marketing agent sending messages in your brand's voice, it's exactly the trust problem I flagged in the Birdsview piece. The capability curve is racing ahead of the control curve, and for marketers, control is the part that determines whether you dare to deploy. The bigger picture: nine of the past twelve months, the size ceiling for open models was set by Kimi. The gap between open and closed keeps shrinking, the cost keeps dropping, and the winners in martech will be the ones who built their data house in time to take advantage. If you needed one more reason to stop paying the double spend on tools and start investing in your foundation, this is it. :) \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Fora raises $60M at a $1 billion valuation, betting that AI makes travel advisors more valuable, not obsolete URL: https://www.martechnext.com/fora-raises-60m-at-a-1-billion-valuation-betting-that-ai-makes-travel-advisors-more-valuable-not-obsolete/ Last updated: 2026-07-17T09:21:20.000Z The travel advisor is back, and now it's a unicorn business. Fora, the New York-based platform that turns career switchers into travel entrepreneurs, has closed a $60 million Series D at a $1 billion post-money valuation. ## The details The round is led by Forerunner and Tactile Ventures, with continued participation from Thrive Capital, Insight Partners, and Heartcore Capital. New names on the cap table include PLUS Capital (alongside Amy Schumer and other members of its artist and athlete collective), BlackPines Capital Partners, and Tribeca Venture Partners. Total funding now stands at $138.5 million. The growth numbers behind the round tell their own story. Fora advisors have booked more than $3 billion in travel since the company launched in 2021\. The first billion took three years. The second took eight months. The third took five. That's the kind of compounding curve investors write checks for. The platform now counts over 15,000 active advisors across 180+ countries, and remarkably, 97% of them are new to the profession: former physicians, attorneys, traders, full-time parents, and retirees. Some advise for supplemental income, others run businesses booking more than $10 million per year. The new capital goes toward deepening Via, Fora's embedded AI assistant currently in beta with a group of top advisors, plus expansion into cruise, flights, enterprise, and new markets. ## The MartechNext take I'll be honest: this is my favorite kind of funding news, because the thesis behind it is contrarian in exactly the right way. The dominant narrative says AI will disintermediate the travel advisor. Why pay a human when ChatGPT can plan your Tuscany trip in thirty seconds? Fora is betting the opposite: AI handles the operational layer (destination research, supplier knowledge, itineraries, proposal generation), so the human can focus on what co-founder Evan Frank calls "human expertise, relationships, taste". That division of labor is precisely how I believe AI should be deployed. Automate the repetitive and predictable first, then let humans do the judgment work that no algorithm replicates. But the more interesting layer, and the reason this belongs in a martech publication, is what Fora actually is underneath: marketing and sales infrastructure for a distributed workforce. Strip away the travel romance and you see a platform that gives 15,000 micro-businesses a booking system, supplier relationships, training, and now an AI operating layer. Every advisor interaction feeds the platform. Every booking makes Via smarter. That's a data flywheel that gets harder to copy every month it runs, and it's the same compounding advantage I keep pointing to in my writing on AI-driven growth. There's also a retention story hiding in plain sight. Travel's structural problem is that acquisition costs keep rising while loyalty stays scarce; most brands are stuck buying the same customers over and over. A trusted human advisor is arguably the most effective retention mechanism the industry has ever had. Nobody churns on the person who saved their honeymoon. If Fora's model works, it's not because AI plans better trips, it's because human relationships quietly fix travel's LTV problem while AI keeps the cost of serving those relationships low. Two things I'd watch. First, advisor churn: when 97% of your workforce is new to the profession and many started as a side hustle, keeping them active and productive is the real operational challenge, and it's exactly what Via needs to prove. Second, the competitive framing. Investor claims about being bigger than "all of AI travel" make for great quotes, but the actual competition isn't other AI travel startups. It's the DIY traveler with a free chatbot. Fora's bet only pays off if human-plus-AI consistently beats AI-alone on outcomes travelers can feel. My money says it will, at least for the complex, high-stakes trips where travel's long and emotional customer journey does its worst. For a $49 city break, nobody needs an advisor. For a three-week multi-generational trip through Asia? That's exactly where automation with human accountability wins. :) \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### WhatsApp Marketing in 2026: what marketers need to know to use it effectively URL: https://www.martechnext.com/whatsapp-marketing-in-2026-what-marketers-need-to-know-to-use-it-effectively/ Last updated: 2026-07-17T09:21:03.000Z From Status Ads and AI chatbots to conversational commerce: WhatsApp is growing into a full-fledged marketing channel. These are the opportunities, the facts, and how to prepare now, wherever you're based. ## WhatsApp is no longer a chat app. It's a marketing channel. Imagine reaching your customer not through an overflowing inbox or a busy social media feed, but in the app they open dozens of times a day to talk to friends and family. The same channel where messages have a 98% open rate, where replies come within minutes, and where conversion works fundamentally differently than on any other digital channel. That channel has existed for a long time. But until recently, for advertisers it was mainly a place for customer service and order confirmations. That is now changing fast. In 2026, WhatsApp is evolving into a full marketing, sales, and advertising platform. And for businesses in strong WhatsApp markets, that opens up new opportunities. One important note before we start: WhatsApp's feature rollout is not uniform. What's live in Brazil or India may still be pending in the EU, and what's available in Germany may not yet be available in the Netherlands. In this article I'll walk you through the most important developments, flag where availability differs per market, and show how you can build an advantage today regardless of where you operate. ## How big is WhatsApp really? Bigger than you think WhatsApp is the most used messaging app in the world, and in many markets it's the most used app, period. It dominates daily communication across Latin America, India, large parts of Europe, Africa, and the Middle East. (The main exceptions are the US, where iMessage and SMS still lead, and markets like China, Japan, and South Korea with their own dominant messengers.) To make it concrete, take the Netherlands, one of Europe's strongest WhatsApp markets. According to Newcom's national social media survey for 2026, WhatsApp counts 13.8 million users there, 12.1 million of whom are active daily. That makes it by far the country's biggest platform, bigger than YouTube, Instagram, and Facebook. For comparison: Instagram has 9.6 million users, TikTok 4.7 million, and LinkedIn 6.5 million. In markets like Brazil, India, and Spain, the picture is similar or even more pronounced. What makes this extra interesting for marketers: WhatsApp is not perceived as social media. People experience it as a personal communication tool, which means messages get far more attention than a post in a feed. That personal context translates directly into performance. Where email struggles with average open rates of 20-25%, WhatsApp consistently reaches 90-98%. Click-through rates sit above 50%, while email is happy with 1-4%. And the response rate on WhatsApp messages exceeds 55%, versus 28% for SMS. WhatsApp combines the reach of a mass medium with the intimacy of a personal conversation. That's what makes it unique for marketers. ## Status Ads: WhatsApp's first true ad format is rolling out The biggest development right now is the rollout of WhatsApp Status Ads and Promoted Channels. After the global launch in June 2025, the European Union was initially excluded because of GDPR review by the Irish Data Protection Commission. In March 2026, Meta confirmed via its DMA compliance report that both ad formats are coming to the EU. Status Ads are now active globally, and they're already appearing for users in several European countries. The rollout is happening market by market, so check the current status for your country before you plan campaigns. (In the Netherlands, for example, advertisers can't run Status Ads yet at the time of writing.) This is a big change. Until now, WhatsApp was an entirely organic or click-to-chat channel in Europe. Once Status Ads become available in your market, programmatic and direct-response placements open up for the first time inside an app that a majority of your customers use daily. The lesson from every previous rollout: the companies that have their strategy and technical foundation ready are the first to profit. ### How do WhatsApp Status Ads work? Status Ads appear as fullscreen ads in the Updates tab of WhatsApp, the place where users view their contacts' status updates. The format is comparable to Instagram Stories Ads, but shown in a considerably more personal context. It works as follows: - The user opens the Updates tab to view status updates from contacts - Between the organic status updates, a fullscreen ad appears (image or video, maximum 30 seconds) - The ad contains a CTA button that directly opens a WhatsApp conversation with the business - For 72 hours after the first interaction, unlimited follow-up messages can be sent at no extra cost In addition, there are Promoted Channels: channel listings in the channels overview of the Updates tab, which businesses can use to grow followers and engagement. Think of a travel company sharing weekly destination tips, or an e-commerce brand publishing exclusive deals for followers. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/69b57ae3db6fe979352244e5_WA_PROMOTED_CHANNELS_STATUS_ADS_BUSINESS_FEATURE_ANDROID.webp) Promoted Channels & Status Ads ### Targeting: privacy-first, but effective WhatsApp Status Ads deliberately use limited targeting to respect user privacy. The available options are location (country, city, or region), device language setting, age category, and activity in the Updates tab. When a user has linked WhatsApp to the Meta Accounts Center, ad preferences from Facebook and Instagram can optionally be included for retargeting and lookalike audiences. Important: personal messages, conversations, group memberships, and contact lists are not used for ad targeting. All private chats remain end-to-end encrypted. In the EU, Meta has introduced a "pay or consent" model: users choose between free with ads, or a paid subscription of roughly 3-4 euros per month to stay ad-free. ### The new KPIs: from clicks to conversations (CPSC) With WhatsApp Status Ads, the primary KPI shifts from clicks to started conversations. This is a fundamentally different kind of measurement than traditional display campaigns. The key metrics become Cost Per Started Conversation (CPSC), the qualification ratio of conversations, the conversion rate of qualified leads, and the response rate on your chat flows. It pays to think about this now, so your measurement and dashboards are ready the moment Status Ads go live in your market. ## WhatsApp marketing goes far beyond advertising Status Ads are only part of the story, and in some markets a part that's still on its way. But WhatsApp has already grown into a complete marketing and sales platform in 2026, with capabilities you can use today. The real power sits in the combination of reach, personalization, and conversational commerce, and you don't need Status Ads for that yet. ### 1\. Conversational commerce: the entire customer journey shifts into a chat Consumers increasingly don't want to browse websites, fill out forms, or wait to be called back. They want to discover products, ask questions, and complete purchases within a single conversation. WhatsApp makes that possible with a product catalog in the chat, in-chat shopping carts, and integrated payment options (availability of payments varies per market). This approach shortens the sales cycle considerably. Instead of the traditional path from ad to landing page to form to follow-up, it becomes: ad, click, conversation, conversion. Fewer steps, less drop-off, higher conversion. ### 2\. AI-driven conversations replace static chatbots The first generation of WhatsApp chatbots worked with standard rule-based flows that regularly frustrated users. More and more companies are now building AI assistants that understand intent, context, and sentiment. These assistants answer complex customer questions, automatically qualify leads based on conversation data, and hand chats over to customer service at the right moment. In travel, for example, this looks like an AI assistant that picks up the conversation after an ad click, asks qualifying questions, gives personalized recommendations powered by a recommendation engine, and guides the customer toward a booking. The difference with the classic approach: instead of stopping at generating a click, you make sure that click actually produces value and converts. ### 3\. From generic blasts to personalized chat flows powered by your first-party data Generic WhatsApp messages will lose their effectiveness over time. Customers increasingly expect contextual communication that takes their history and preferences into account. By connecting WhatsApp to your CRM or CDP, in combination with an AI model, you can trigger behavior-driven messages, adapt campaigns based on browsing and purchase history, and keep conversations consistent across teams. In practice, this means connecting your WhatsApp flows to systems like Klaviyo, HubSpot, or your CDP, and using a recommendation engine for the right predictive suggestion. That way WhatsApp doesn't become an isolated channel, but an integrated part of your complete marketing ecosystem, where email, social, website, and WhatsApp reinforce each other. ### 4\. Click-to-WhatsApp Ads via Facebook and Instagram Separate from the new Status Ads, Click-to-WhatsApp Ads via Facebook and Instagram have been available for a while and are proven effective, in every market where WhatsApp is used. These ads appear in the feed or stories of Facebook and Instagram and open a WhatsApp conversation directly on click. The advantage: you combine the advanced targeting of Meta's ad platform with the directness of a WhatsApp conversation. The complete targeting options of Meta Ads Manager are available: demographic targeting, geographic targeting, behavioral and interest targeting, custom audiences, and lookalike audiences. This makes Click-to-WhatsApp Ads particularly suitable for lead generation in sectors where the personal conversation plays a crucial role, such as travel, real estate, education, and financial services. ### 5\. WhatsApp Business API: essential for serious applications All professional WhatsApp marketing activities run on the WhatsApp Business API. The standard WhatsApp Business App is fine for small businesses, but offers limited functionality: broadcast lists are capped at 256 contacts, there's no automation, and no clickable buttons. The API offers scalability, automation, template approval, and CRM/CDP integration. Meta uses a conversation-based pricing model: a "conversation" is a 24-hour window within which unlimited messages can be exchanged for a fixed rate. Costs vary per country and conversation type, with marketing conversations being the most expensive, followed by utility, authentication, and service. Note that WhatsApp qualifies almost everything as a marketing conversation; in Western European markets, a 24-hour window costs around 13 cents, but check the rate card for your countries. ## Practical applications: how you can use WhatsApp today The theory sounds promising, but how do you translate this concretely into your marketing strategy? Here are some example use cases. ### E-commerce Follow up on abandoned carts via WhatsApp instead of email. Send product recommendations based on browsing behavior. Run abandoned cart series with AI-personalized suggestions. Automate post-purchase follow-up with reviews, upsells, and repeat purchases. The higher open rates translate directly into higher recovery rates. ### Travel The travel industry is exceptionally well suited to WhatsApp marketing. Once Status Ads are available in your market, a last-minute campaign via a Status Ad with an urgent CTA becomes possible: interested travelers are connected directly to an AI assistant that checks availability and guides the booking. Seasonal early-bird campaigns combine inspiring destination videos with personalized advice. And a WhatsApp channel with weekly destination tips builds an engaged audience today for future campaigns. With Click-to-WhatsApp Ads via Facebook and Instagram, you can start right now. ### B2B and services Lead qualification via conversational funnels: instead of a form on your website, a prospect starts a WhatsApp conversation in which an AI assistant asks the right questions and routes directly to the right account manager. Quote requests, demo sign-ups, and questions are handled more personally and faster. ## GDPR and privacy compliance: practical rules to stick to In the EU, strict rules apply to WhatsApp marketing, and similar consent-based rules exist in many other markets. This is not a side issue but a core part of your strategy. The most important rules: - **Opt-in is mandatory.** You need explicit opt-in consent: a pre-ticked box in your terms and conditions doesn't count. You need a clear, affirmative action, and you must document it. - **Unsubscribing must be easy.** Every marketing message must contain a clear opt-out option. - **Respect frequency.** Too many spam complaints lower your phone number quality score and limit how many messages you can send per day. - Start with a maximum of 1-2 messages per week and adjust based on engagement. ## Your foundation for a WhatsApp strategy in 2026: a first roadmap The companies that will get the most out of WhatsApp marketing are the ones doing their homework now. Here's a concrete roadmap to get started quickly. **Step 1: Get your technical foundation in order.** Make sure you have a WhatsApp Business Account (or the API for serious use cases), that your Facebook Business Page is linked to your WhatsApp number via Meta Business Manager, and that you have access to Meta Ads Manager. **Step 2: Define your channel strategy.** WhatsApp doesn't replace channels, it strengthens them. Determine where WhatsApp has the biggest impact in your customer journey. Is that lead generation? Post-purchase follow-up? Customer service? Start with one clear use case and build from there. **Step 3: Build conversational content.** WhatsApp demands a different creative approach than display or search ads. Messages that feel like advice perform better than pushy sales pitches. Develop content that fits a conversational context, and map out which questions your customers ask at every stage of the customer journey. (You should be doing this anyway for your GEO optimization.) **Step 4: Invest in automation.** Manually replying to every WhatsApp message doesn't scale. Invest in AI-driven chat flows that qualify leads, answer questions, and route to the right team. The combination of an AI assistant with human escalation is the sweet spot. **Step 5: Measure, learn, optimize.** Don't steer on opens and clicks alone, but on conversation quality, conversion rate, and customer value. The data from WhatsApp conversations is gold for improving your targeting, your offering, and your complete marketing ecosystem. ## What a modern WhatsApp marketing stack looks like Using WhatsApp as a marketing channel sounds simple, but there's more to it than creating a Business Account and sending some messages. You need the right technical foundation, smart automation, and most importantly: the data that makes every message relevant. The building blocks: **A recommendation engine as the intelligence layer.** The quiet force behind effective WhatsApp marketing is a system that analyzes customer behavior, purchase history, and profile data to predict which products, destinations, or offers best fit each individual customer. That intelligence shouldn't live in WhatsApp alone: the same layer should power your website personalization, email campaigns, and ads, so every channel gets smarter. **An AI assistant trained on your business data.** Not a generic chatbot with canned answers, but an assistant trained on your product range, availability, prices, terms, and frequently asked questions. The result is an assistant that answers as if your best employee is on the line, 24/7 and at scale. This matters most in industries where the advisory conversation is crucial to the purchase, such as travel. **1-on-1 WhatsApp newsletters with AI follow-up.** A fundamentally different model than the traditional email newsletter: instead of one message to an entire list, you send individual messages tailored to each recipient's profile and preferences, with an AI assistant behind them that responds instantly when a customer replies. A concrete example: an e-commerce company sends an offer for a specific clothing item via WhatsApp. The consumer replies that it's not their preferred color or style. Instead of a dead end, the AI assistant picks up the conversation. Based on the reply, previous purchase behavior, and what similar customer profiles bought, the assistant immediately makes new, relevant recommendations. What started as a simple newsletter becomes a personalized sales conversation, fully automated, and turns into a genuine additional sales channel. This works in any industry. A travel company sends summer holiday inspiration. The customer replies that their budget is lower than the offers shown. The AI assistant adjusts on the spot and presents alternatives that fit the budget, the travel period, and the preferences from earlier interactions. Every conversation generates data that makes the next recommendation even better. **No silo, but an accelerator for your entire customer journey.** The most important principle: WhatsApp should not become a standalone channel. The data from WhatsApp conversations, profiles, and interactions should flow back into your central intelligence layer and be used across your whole marketing ecosystem. A preference someone shares via WhatsApp improves the product recommendations on your website, the personalization in your emails, and the targeting of your ads. Everything reinforces everything else. ## Conclusion: WhatsApp is becoming a major channel, so get in early WhatsApp marketing is making a giant leap in 2026\. The combination of massive daily reach, the ongoing rollout of Status Ads across markets, increasingly powerful AI assistants, and the shift toward conversational commerce creates an enormous opportunity. The history of digital advertising always shows the same pattern: early movers win. The companies that mastered Google Ads (AdWords) and Facebook Ads first profited for years from lower costs and less competition. The same was true for Google Shopping, Instagram, and TikTok. WhatsApp is next. The difference with earlier channels is that WhatsApp adds an extra layer: the conversation. Not just reach, not just a click, but an actual interaction with your potential customer. While Status Ads may still be on their way in your market, you can build a head start today with Click-to-WhatsApp Ads, the Business API, AI chat flows, and CRM/CDP integration. Whoever has their foundation in order now will be first in line when Status Ads go live. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### The power of intelligent personalization: How AI will transform the customer journey URL: https://www.martechnext.com/the-power-of-intelligent-personalization-how-ai-will-transform-the-customer-journey/ Last updated: 2026-07-17T09:20:49.000Z **In this article:** - Why the complex travel funnel demands intelligent personalization - The difference between traditional SVD and modern deep learning recommendation engines - The multiplier effect: how a better customer experience leads to sustainable growth - The future: AI agents and real-time guidance throughout the complete customer journey The travel industry is known for its complex customer journey. Unlike many other sectors where customers make a purchase within minutes or hours, the average time to booking in travel can stretch to weeks or even months. This unique dynamic creates both challenges and opportunities for (travel) organizations aiming for the best possible customer experience and conversion. Over the past years, I've experienced up close how technology can transform this industry. What started as a fascination with the complexity of traveler behavior grew into a mission to help companies create genuinely relevant, personal experiences. In this article, I share my vision on where this technology stands today, where it's heading, and why I believe we're on the eve of a revolution in customer experience, driven by connecting data properly. I'll use the travel industry as my example, but this applies to any industry with a meaningful customer journey. ## The complexity of the travel funnel: a challenge of its own The booking process for a trip is fundamentally different from other online purchases. Where consumers often buy a book or clothing on impulse, travelers go through an extensive orientation phase spanning multiple touchpoints, channels, and decision moments. Think about the typical journey of a potential vacationer: it might start with an Instagram photo of a friend on a tropical beach. This triggers the first research: which destinations are out there? What's the best season? Then the comparing begins: accommodations, flights, car rental. In between, reviews are read, travel blogs consulted, and price comparison sites visited. There are discussions with travel companions: can the kids come along? Does it fit everyone's calendar? Does it fit the budget? ``` ๐Ÿ“ธ Instagram inspiration โ†“ ๐ŸŒ Destination research โ†“ ๐Ÿจ Comparing hotels โ†’ โœˆ๏ธ Checking flights โ†’ ๐Ÿš— Car rental โ†“ โญ Reading reviews โ†’ ๐Ÿ’ฐ Comparing prices โ†“ ๐Ÿ‘จโ€๐Ÿ‘ฉโ€๐Ÿ‘งโ€๐Ÿ‘ฆ Discussing with travel companions โ†“ ๐Ÿ“… Checking dates โ†’ ๐Ÿ’ณ Booking ``` This complexity is amplified by external factors: - ๐ŸŽ„ **Seasonal booking patterns**: Long-haul trips are booked months in advance, city breaks often last-minute - ๐Ÿ“ˆ **Dynamic availability**: Hotels fill up, flights get more expensive as the date approaches - ๐Ÿ’ **Emotional decision-making**: A vacation is often the biggest expense of the year - ๐Ÿ‘ฅ **Multi-stakeholder decisions**: Different family members with different wishes and preferences ๐Ÿ’ก *"A customer journey that stretches over weeks or months, with dozens of interaction moments spread across different channels."* The result? A customer journey that stretches over weeks or months, with dozens of interaction moments spread across different channels. And this is exactly where the problem lies: most travel organizations treat every contact moment as an isolated event. The visitor looking at Italy today still gets generic offers for every possible destination tomorrow. We analyzed this challenge together with Bookit, known for brands like Weekendjeweg.nl and Traveldeal. We started with a simple hypothesis: what if we simply adapted content and visuals to broad, well-known audience segments? Not even one-to-one personalization, just different messages for different groups. We identified three broad audiences, inspiration seekers, practical planners, and luxury lovers, and created tailored content and imagery for each segment. Where inspiration seekers saw dreamy images of undiscovered destinations, practical planners got concrete information about facilities and prices. Luxury lovers saw exclusive experiences and premium accommodations. ``` ๐Ÿ‘จโ€๐Ÿ‘ฉโ€๐Ÿ‘ง Families โ†’ ๐ŸŽ  Kids' activities + ๐ŸŠ All-inclusive ๐Ÿ’‘ Couples โ†’ ๐Ÿท Romantic dinners + ๐ŸŒ… Boutique hotels ๐Ÿ‘ฅ Friends โ†’ ๐ŸŽ‰ Nightlife + ๐Ÿ„ Adventure ``` The results were already excellent. This simple form of segmentation alone led to significant improvements in conversion and engagement. No surprise, of course: anyone can figure out that something converts better when it matches your exact wishes. The really interesting question is how to take this segmentation to true one-to-one personalization. Not by creating audience segments upfront and trying to fit people into a box, but by presenting tailored offers across channels based on data points from the entire customer journey. (Think website, email, social media, and so on.) ``` ๐Ÿ“ฑ Mobile โ†’ ๐Ÿ’ป Desktop โ†’ ๐Ÿ“ง Email โ†’ ๐Ÿ“ฑ Mobile โ†’ ๐Ÿ›’ Booking ``` This is the problem I've spent the past years solving: truly personalizing with good recommendations, and being present throughout the entire journey with relevant contact moments. We started with the "standard route": implementing a Customer Data Platform to make some simple changes to the website and email communications based on recognition. We later expanded this with a recommendation engine that could make suggestions based on machine learning. That already delivered very good results. But we ran into its limits, both in intelligence and in the cost of running the model as data volumes grew. That was the moment, partly driven by the rapid developments in AI, to build our own recommendation engine. The starting points: it had to be self-learning, fast, and cost-efficient. I've written about it before, but I enjoy sharing what makes it different from traditional models, the practical applications (and results achieved), and how it's now woven into the entire customer journey, from searching for a trip, to the experience during the trip, to inspiration for the next destination afterwards. ## From linear models to deep learning: the evolution of recommendation engines Traditional recommendation engines in travel often work with Singular Value Decomposition (SVD) models. SVD is essentially a mathematical technique that reduces large matrices of user preferences to smaller, manageable dimensions. Think of it as an enormous spreadsheet with millions of rows (users) and columns (trips) being "compressed" into a workable model. Although SVD models were the standard for years, they suffer from fundamental limitations (for this article I'm deliberately simplifying some concepts, the reality is of course more nuanced): - โŒ **Linear assumptions**: SVD assumes preferences are linear. If you rate Spain an 8 and France a 6, then Portugal should be a 7\. But human behavior doesn't work that way. - โŒ **Cold start problem**: New users without history are a black box for SVD - โŒ **Static representations**: Once trained, SVD models don't adapt to changing preferences - โŒ **No understanding of context**: A romantic city trip is treated the same as a weekend away with friends The deep learning recommendation engine we developed takes advantage of what's now possible and approaches the math from a different angle. Instead of linear mathematical reductions, it uses Neural Collaborative Filtering, an architecture that can actually model the complexity of human behavior. The advantages sit mainly in: - โœ… **Self-learning and adaptive**: Where SVD models need to be fully retrained periodically (a (very) costly and time-consuming process), the deep learning model learns continuously. Every interaction makes the system smarter, without recalculating the entire model. - โœ… **Intelligent pattern recognition**: The system spots subtle connections that SVD misses. It understands, for example, that someone who first looks at budget destinations and then suddenly browses luxury resorts is probably saving up for something special. Not random behavior, but a pattern with meaning. - โœ… **Cost-efficient**: Because the model learns incrementally instead of being fully retrained, operational costs are significantly lower. On top of that, better targeting means you waste less expensive marketing money. - โœ… **Real-time performance**: While SVD models are often updated in overnight batches, this engine generates real-time recommendations that respond directly to the user's current behavior, immediately applicable across channels. The crucial difference? Where SVD models forcibly try to squeeze complex human preferences into a linear mathematical model, deep learning embraces the complexity. The system understands that the journey from inspiration to booking isn't linear, but a complex web of emotions, practical considerations, and external factors. Another crucial advantage of this deep learning approach is its self-learning character. Every interaction, from website visit to booking, feeds the model with new data. This creates a positive feedback loop where: - ๐Ÿ“ˆ Predictions become increasingly accurate - ๐Ÿ” New trends are identified automatically - ๐Ÿ—“๏ธ Seasonal patterns refine themselves year after year - ๐Ÿ‘ค Individual customer profiles get richer The system also adapts to changing market conditions. Certain destinations or trip types can suddenly be on the rise. Where you'd normally spot this in after-the-fact analysis, the model notices increased demand from interactions (even before the booking) and can adjust recommendations in real time. This makes it possible to predict not only what a customer wants, but also why, and more importantly (and for me the holy grail): what the next step in their journey will be. This self-learning aspect reinforces the multiplier effect: the longer a customer interacts with the platform, the better the recommendations get, and the higher the chance of repeat purchases. ๐Ÿ’ก **The multiplier effect: the longer a customer interacts with the platform, the better the recommendations get, the higher the chance of repeat purchases.** ## Intelligent business rules: domain knowledge combined with AI What makes this recommendation engine unique is the integration of domain-specific knowledge into the AI model. The system takes into account factors that are crucial for the travel industry. Some examples: ### ๐ŸŒค๏ธ Weather-dependent recommendations ``` ๐ŸŒง๏ธ Raining in NL โ†’ ๐Ÿค– AI detects โ†’ โ˜€๏ธ Promote sunny destinations ``` When it rains in the Netherlands, the system automatically promotes sunny destinations. When the weather at home is beautiful, recommendations shift toward domestic weekend getaways. This contextual intelligence not only increases relevance, but also creates a feeling of "they get me" with the customer, the foundation for long-term customer relationships. ### ๐Ÿ“… Seasonal predictions ``` ๐ŸŽ„ December: Christmas markets booked (in July) โ˜€๏ธ Summer: Last-minute city trips ๐ŸŽฟ Winter: Planning long-haul trips ๐ŸŒธ Spring: Locking in summer holidays ``` The system "knows" that Christmas market trips are booked as early as July, that families travel during school holidays, and that last-minute deals mainly attract solo travelers. These insights are automatically weighed into the recommendations, so customers get the right suggestions at the right moment. ### โฐ Availability and urgency ``` โฐ "2 days left!" โ†’ 2๏ธโƒฃx boost in ranking ๐Ÿ”ฅ "Almost sold out" โ†’ ๐Ÿ“ˆ Higher priority โœ… Available โ†’ ๐Ÿ‘๏ธ Show โŒ Full โ†’ ๐Ÿšซ Auto-filter ``` Offers that are about to expire get a 2x boost factor in the ranking. The system automatically filters out unavailable options and takes the minimum booking window per trip into account. This prevents frustration and increases the chance of actual bookings. ### ๐ŸŽจ Diversity in recommendations To prevent monotony, the system limits the number of recommendations per city. For thematic recommendations, such as Formula 1 trips, overlap between similar items is intelligently minimized. This keeps the offering varied and keeps inspiring customers for future trips. ## Practical examples: personalization in action The power of the recommendation engine comes to life in concrete applications across all the channels where the consumer actually is: ### ๐Ÿ’ป Website personalization On product detail pages (PDPs), visitors see contextual recommendations that match their interest. A family looking at an all-inclusive resort in Turkey gets similar family-friendly options, while a couple browsing luxury city trips is shown other romantic destinations. The copy is dynamically rewritten based on the travel party profile, which at Bookit resulted in a 20% higher PDP-to-checkout conversion, for example. ### ๐Ÿ“ง Intelligent newsletters Instead of generic newsletters, subscribers receive personalized content based on their position in the customer journey. "Browsers" get inspiring content featuring multiple destinations, while "ready-to-book" profiles receive concrete deals with booking incentives. The system prevents saturation by excluding previously sent recommendations for X days and boosting recently viewed deals for X days. This intelligent approach makes customers look forward to the next newsletter instead of unsubscribing. A nice side effect: your deliverability becomes (and stays) excellent. ### ๐Ÿ“ฑ Social media targeting The recommendation engine also drives social campaigns. Instead of advertising broadly, the system creates targeted campaigns per audience and funnel stage. Videos with "3-city inspiration" reach upper-funnel prospects, while concrete deals with AI-generated USPs are shown to visitors further along in their journey. As an example, this orchestrated approach resulted in 23% more add-to-cart actions and lower acquisition costs. Part of that freed-up budget can then be reinvested in attracting new customers. (More on that later.) ### ๐Ÿ’ฌ Conversational commerce An innovative module that works hand in hand with the recommendation engine is a travel inspiration wizard: an interactive tool that generates personalized travel brochures based on the user's answers. By connecting the recommendation engine to conversational interfaces like this, travelers can discover their perfect trip through natural interaction. Meanwhile, every interaction enriches the customer profile, making future recommendations even more relevant. ``` ๐Ÿ“Š Data collection โ†’ ๐Ÿง  AI processing โ†’ ๐ŸŽฏ Personalization โ†’ ๐Ÿ“ฑ Multi-channel delivery ``` ## The Bookit business case: numbers and results An important aspect of a recommendation engine is that it lets you identify profiles. Once you can do that, you have a group of profiles with already elevated interest. That nuance needs to be made, but even so, the results for people who received personalized experiences are impressive: - ๐Ÿ“ˆ Newsletter: 5.7x higher conversion compared to non-personalized visitors - ๐Ÿ’ฐ Newsletter: 6.3x higher revenue per user - ๐Ÿ›’ 23% more add-to-cart actions through intelligent multi-channel targeting - ๐ŸŽฏ 41% higher conversion on specific A/B tests for homepage personalization - ๐Ÿ‘† 12.5% higher book-now click rate (73% vs. 65% in the control group) These numbers are obviously great, but the real power lies in what happens next: the multiplier effect that amplifies these initial successes. ## The multiplier effect: from conversion to customer value The impact of intelligent personalization reaches far beyond direct conversion. By offering customers a more relevant experience, a powerful flywheel effect emerges that can transform the entire business. (Or put simply: your LTV:CAC ratio becomes better than everyone else's.) ``` ๐Ÿ“ˆ Better experience โ†“ ๐Ÿ’š Higher loyalty โ†’ ๐Ÿ”„ More frequent returns โ†“ โ†“ ๐Ÿ’ฐ Higher CLV โ†’ ๐Ÿ“ข Word-of-mouth โ†“ โ†“ ๐Ÿš€ More budget for acquisition โ† ๐Ÿ‘ฅ New customers ``` ### ๐Ÿ’š Increased customer loyalty When travelers notice that a platform truly "gets" them and makes relevant suggestions, trust grows. They come back sooner for their next trip, which significantly increases customer lifetime value. At Bookit, for example, we saw that personalized customers returned 2.4x more often within 90 days. ### ๐Ÿ’ธ More efficient marketing spend With higher customer value, travel organizations can invest more in acquiring new customers. Improved targeting also ensures marketing budgets are spent more efficiently (simple example: no more ad budget flowing to existing customers who would have booked anyway). That freed-up budget can be invested directly in growth. ### ๐Ÿ† Competitive advantage In a market dominated by price comparison sites, where loyalty is scarce, superior personalization offers a sustainable competitive advantage. It becomes the reason customers come back, even when they could be a few euros cheaper elsewhere. The emotional bond created by relevant, personal experiences is hard for competitors to copy. ### ๐ŸŒ Network effects Satisfied customers share their positive experiences, which leads to organic growth. Every new customer who arrives through a recommendation already has positive expectations and is more likely to become a loyal customer themselves. This amplifies the multiplier effect exponentially. ## The future: AI agents and presence throughout the complete journey ๐Ÿš€ The next step is combining the recommendation engine with AI agents. This integration creates unprecedented possibilities across the complete customer journey. This is exactly the part that gets me genuinely excited. :) A few examples of what's being built right now: ### ๐Ÿค– Intelligent chatbots with context Imagine: a customer chats with an AI assistant about possible destinations. Instead of generic answers, the chatbot has direct access to the full customer profile through the recommendation engine. The bot "knows" this customer has been to Spain before, prefers family vacations, and usually books during the May school break. The result? Hyper-personalized advice that feels like a conversation with a personal travel advisor, and lays the foundation for a long-term customer relationship. ### ๐ŸŽฏ Proactive customer service When a customer contacts customer service, the AI agent already has context. Is the customer leaving within a week? Then the agent can proactively help with last-minute questions. Has the customer recently looked at similar destinations? The agent can immediately offer relevant alternatives in case of changes or cancellations. ### ๐Ÿ’ฌ Real-time travel guidance via WhatsApp An exciting development is WhatsApp journey technology, which brings personalization to the missing moment: during the trip itself. This kind of assistant elevates the customer experience with, for example: - ๐Ÿ–๏ธ **Proactive tips at the right moment**: "The weather is perfect for a beach visit today. Here are three beach clubs that match your preferences." - ๐ŸŽซ **Personalized vouchers and tickets**: Specific discounts or personalized suggestions for entrance tickets, for a particular museum, for example. - ๐ŸŽฏ **Smart activity suggestions**: Based on the travel party, combined with that day's weather and historical behavior. - ๐Ÿ“ฒ **Direct ticket purchases**: Bookable directly via WhatsApp, without leaving the app. ``` ๐Ÿ“ Location check โ†’ ๐ŸŒค๏ธ Weather analysis โ†’ ๐Ÿ’ก Suggestion โ†’ ๐ŸŽŸ๏ธ Directly bookable ``` The interesting part is that all these interactions contribute to one central customer profile. Every WhatsApp conversation, every voucher used, every activity booked enriches the profile. When the customer returns for the next booking, the system has become even smarter. The multiplier effect in action: every action makes the next experience better. ## The flywheel effect of integrated AI This combination of a recommendation engine and AI agents creates an incredibly powerful flywheel: ``` Pre-trip: ๐Ÿ’ฌ Chatbots โ†’ inspiration & planning โ†“ โœˆ๏ธ Booking: ๐ŸŽฏ Personalized suggestions โ†“ ๐Ÿ–๏ธ On the trip: ๐Ÿ’ฌ WhatsApp assistant โ†’ relevant tips โ†“ ๐Ÿ  After the trip: ๐Ÿค– Agents โ†’ preparing the next trip โ†“ ๐Ÿ”„ Repeat, with even better personalization ``` - **Before the trip**: Chatbots help with inspiration and planning, fed by recommendation data - **During booking**: Personalized suggestions and dynamic content increase conversion - **On the trip**: A WhatsApp assistant keeps customers engaged with relevant tips and offers - **After the trip**: Follow-up via agents that are already preparing the next trip Every phase strengthens the next, every interaction makes the system smarter, and every satisfied customer increases the chance of repeat purchases and recommendations. ## Conclusion: the journey toward truly great customer experience through personalization is underway The travel industry is at a crossroads. On one side, technological developments make unprecedented personalization possible. On the other, consumers expect ever more relevance and convenience. Organizations that seize this opportunity and invest in intelligent recommendation engines position themselves well for the long term. Especially in an era of rising advertising costs, shrinking visibility, and heavy competition. ``` ๐Ÿšฆ CROSSROADS | Old road ๐Ÿ›ค๏ธ โ†โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ†’ ๐Ÿš€ New road | โŒ Generic โœ… Personalized โŒ Reactive โœ… Proactive โŒ Fragmented โœ… Integrated ``` In an industry where margins are under pressure and acquisition costs keep rising, intelligent personalization can be the difference between surviving and thriving. It's no longer about optimizing individual touchpoints, but about creating an integrated ecosystem where every interaction adds value. For travel organizations ready to take the next step, it seems clear to me: the technology exists, the business case is proven, and the time to act is now. Because in the world of travel, it's ultimately about creating unforgettable experiences, and that starts at the very first digital contact. But let's be honest: these questions are not limited to the travel industry. Whether you work in retail, financial services, education, or healthcare, wherever customers go through a journey, wherever choices are complex, and wherever personalization can make the difference, these principles apply. The technology developed for travel keeps proving valuable in sectors nobody initially thought of. And that brings me to what I find most fascinating about these developments: the conversations with people facing similar challenges. Whether you're a tech enthusiast working on similar solutions, or an entrepreneur wrestling with customer experience in your own industry, I get energized by exchanging ideas on these topics. How do you translate complex data into human experiences? Where is the line between personalization and privacy? How do you build systems that truly learn and adapt? I don't have all the answers, and probably even AI doesn't ;). The technology is developing at breakneck speed, customer behavior keeps evolving, and what works today can be outdated tomorrow. But that's exactly why the dialogue is so valuable. By sharing experiences, learning from each other, and innovating together, we get further than we would alone. Consider this article an invitation. Curious how these concepts could work in your industry? Working on innovative customer experience solutions yourself? Or simply in the mood to exchange thoughts about the future of personalization? I'd love to hear from you. Because the best ideas often emerge from the cross-pollination between different worlds. ๐Ÿ’ก \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Focus on value, not volume! Why e-commerce companies should care more about their LTV/CAC ratio URL: https://www.martechnext.com/focus-on-value-not-volume-why-e-commerce-companies-should-care-more-about-their-ltv-cac-ratio/ Last updated: 2026-07-17T09:20:34.000Z The shift every e-commerce company needs to make to grow sustainably in a world of exploding acquisition costs. ## The question almost nobody asks "How many new customers did we bring in this month?" I hear this question constantly. But it's the wrong question. The question you should be asking is: "How much value did we create per euro invested?" The difference between these two questions is the difference between a company that will grow sustainably and a company that will come under increasing pressure. The reality is that acquisition costs have risen by 222% over the past 10 years, while most companies are still chasing volume instead of value. ## The basics: what are LTV and CAC, and what do they actually tell you? ### Lifetime Value (LTV) Lifetime Value (or simply put: customer value) is the total net profit a customer generates over the entire relationship with your company. (Note: net profit, not revenue. This distinction is often forgotten, but it's crucial.) Let's start with a simplified example of how to calculate LTV. (I'll come back to more detailed and precise calculations later.) A fashion webshop has the following data: - Average order value: โ‚ฌ85 - Purchase frequency: 3.2x per year - Average customer lifespan: 2.3 years - Gross margin after cost of goods: 65% - Operating margin (after fulfillment/shipping): 45% The calculation, step by step: 1. Total revenue per customer = 85 ร— 3.2 (purchase frequency) ร— 2.3 (lifespan) = โ‚ฌ625 2. Gross LTV = 625 ร— 65% (after cost of goods) = โ‚ฌ407 3. Net LTV = 625 ร— 45% (after cost of goods and other costs) = โ‚ฌ282 That's a โ‚ฌ125 gap between gross and net. And that's exactly where many companies go wrong. They calculate with gross LTV but forget things like fulfillment (order picking and packaging), returns, and customer service. ### Customer Acquisition Cost (CAC) CAC (acquisition cost) covers all the costs of bringing in a new customer. Not just your Google Ads and Meta campaigns. Again, a simplified example from the same fashion webshop: Monthly costs: - Google Ads: โ‚ฌ25,000 - Meta Ads: โ‚ฌ20,000 - Influencer spend: โ‚ฌ8,000 - Marketing team (3 FTE): โ‚ฌ15,000 - Tooling (email platform, analytics): โ‚ฌ2,000 - External content creation: โ‚ฌ5,000 - Total: โ‚ฌ75,000 With this spend, the webshop brings in 450 new customers. The acquisition cost per customer is then: 75,000 / 450 = โ‚ฌ167 (CAC). Looks fine when you compare it to the net LTV above, but are all of those customers actually new customers? I'll come back to that! ## Why focusing (only) on new customers is often throwing money away ### The skewed split in marketing budgets I see it constantly at companies: an extreme focus on acquisition costs for (new) customers, per month or per channel. Not entirely illogical when you're focused on growth, and I've certainly been guilty of it myself :). Especially in the first years of a company, it feels like the most important KPI. And yet it's really quite simple. It's between 5 and 25 times more expensive to acquire a new customer than to retain an existing one. If you compare conversion rates of existing customers versus new customers, you'll see what I mean. And still, at most companies, somewhere between 80% and 90% of the marketing budget goes to acquisition. ### Incremental value Incremental value is the value you generate as a direct result of a marketing activity. It's value you would not have generated if you switched that activity off. Put simply: how much of your marketing actually drives extra sales that wouldn't have happened otherwise? A practical example: A home & living webshop was spending โ‚ฌ40,000 per month on Google Ads. Google reported 600 "new customers". That works out to a CAC of โ‚ฌ67\. Not a bad number for that industry. We ran a holdout test (an experimental method where a group is excluded, the "holdout group", to measure the true performance of a channel or the long-term effects of a change). We paused all ads in 25% of the Netherlands (random postal codes). After 4 weeks, it turned out: - 40% of the "new customers" were existing customers coming back through ads - 30% would have found the website organically anyway - Only 30% was genuinely incremental (extra sales driven by the ads) The real acquisition cost then becomes: 40,000 / (600 ร— 30%) = โ‚ฌ222\. Quite a different number. You could even argue (a bit bluntly) that this webshop was throwing away โ‚ฌ28,000 per month (โ‚ฌ336,000 per year), money that would most likely have been far better spent on retention (loyalty, customer service, and so on). ## A simple strategy based on your LTV/CAC ratio Now that we know the basics of LTV and CAC and which variables play a role, you can already build a better strategy for yourself. Below is a simple example of how you could start steering on this. (Later I'll go deeper into how to calculate the ratios more precisely and keep them up to date.) The ratio between customer value and acquisition cost determines whether your business is healthy. This ratio obviously depends on your margins. If you sell a software product, you might have 85-90% margins, while the average e-commerce company has to make do with 40% (or less). Let's assume an average. At a 3:1 ratio, you earn โ‚ฌ3 in customer value for every โ‚ฌ1 you spend on acquisition. That gives you enough room for: - Operating costs (30-40% of revenue) - Product development and innovation (10-15%) - Profit margin for growth investments (15-20%) - A buffer for seasonal fluctuations and market changes So your target is 3:1\. If you track these ratios and see them shift, here are the kinds of actions you could take based on where you currently stand: ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/ltv-cac-ratio-table.png) LTV/CAC Ratio A practical example of a fictional webshop with different ratios over the past 3 years: **Year 1: LTV โ‚ฌ95, CAC โ‚ฌ78, ratio 1.2:1** - Nearly bankrupt. - Stopped all paid acquisition - Focused on improving the offer - Only used owned channels like SEO and email **Year 2: LTV โ‚ฌ135, CAC โ‚ฌ45, ratio 3:1** - Break-even - Carefully restarted advertising - Marketing budget split between retention and (partly) acquisition - Focus on improving LTV **Year 3: LTV โ‚ฌ205, CAC โ‚ฌ58, ratio 3.5:1** - Healthy growth. Acquisition costs rise slightly, but LTV rises faster in proportion - Focus on scaling up advertising for new customers, now that the LTV/CAC ratio is more than healthy It's a simple example, but the lesson is: focus on creating value first (raising LTV), then on volume (new customers). ## What's next? Calculating LTV, from basic to advanced We've used a few simple examples for calculating LTV so far. But LTV is not a fixed number and will keep changing (hopefully in the right direction :)). To calculate and track it properly, you need a fair amount of data (and a bit more technical knowledge). To approach this step by step, I've split the calculation methods into 3 phases. You can read this as a path: whether you're an established webshop that wants to get started, or a younger company that wants to grow into more advanced models as your data grows. (From start-up to scale-up, for example.) ### Phase 1: The quick & dirty method (less than 6 months of historical data) For starters who want to begin today (and as worked out earlier in this article). Take the simplest formula: **LTV = Average order value ร— Frequency ร— Lifespan ร— Margin** A practical example: - Average order: โ‚ฌ75 - Estimated frequency: 2x per year (industry average) - Estimated lifespan: 18 months - Gross margin: 60% LTV = 75 ร— 2 ร— 1.5 ร— 0.60 = โ‚ฌ135 (As an alternative, use the payback period: within how many months do you earn back your acquisition costs? A good target for e-commerce: within 6 months.) ### Phase 2: Cohort-based LTV (6 to 18 months of historical data) Now it gets more interesting. We follow groups of customers (cohorts) over time. A practical example from, say, an online drugstore: Take all customers who made their first purchase in January 2025\. Track their spending per month. ![](https://storage.ghost.io/c/74/6b/746bc641-7bf2-4c8a-b4cf-e4d80230e3c6/content/images/2026/07/cohort-revenue-table.png) Cohort Revenue Total after 6 months: โ‚ฌ106 in revenue per customer. With a 60% margin: LTV = 106 ร— 0.60 = โ‚ฌ64 **The extrapolation method:** We only have 6 months of data, but we want to forecast 24 months. We can see that revenue declines by roughly 15% per month (the "decay rate", an exponential decline over time). The calculation for months 7 to 24: - Month 7: 11 ร— 0.85 = โ‚ฌ9.35 - Month 8: 9.35 ร— 0.85 = โ‚ฌ7.95 - ...and so on Total after 24 months: โ‚ฌ156 With a 60% margin: LTV = 156 ร— 0.60 = โ‚ฌ94 Tip: compare different cohorts. If you compare your February customers with your January customers, you'll often see big differences. Instagram customers might have a 40% higher LTV in January than Google Shopping customers in the same period. Those insights are pure gold for allocating your marketing budget. ### Phase 3: Probabilistic models (18+ months of historical data) For companies that truly want to be data-driven. The BG/NBD model (Beta-Geometric/Negative Binomial Distribution, quite a mouthful, I know) solves a fundamental problem: when is a customer "dead" (churned) versus "dormant" (might come back)? A practical example (I'll spare you the exact math :)): **Customer A, the loyal customer:** - Last purchase: 45 days ago (Recency) - Total number of purchases: 5 (Frequency) - Customer since: 365 days (T) - Average order value: โ‚ฌ120 The model calculates: - Probability the customer is still "alive": 78% (despite 45 days of silence, because 5 purchases in a year is an active customer) - Expected purchases in the coming year: 3.2 - Predicted LTV: 120 ร— 3.2 ร— 0.60 margin = โ‚ฌ230 **Customer B, the one-off shopper:** - Last purchase: 180 days ago - Total number of purchases: 2 - Customer since: 365 days - Average order value: โ‚ฌ95 The model calculates: - Probability the customer is still "alive": 23% (a long silence plus few purchases usually means they're gone) - Expected purchases in the coming year: 0.4 - Predicted LTV: 95 ร— 0.4 ร— 0.60 = โ‚ฌ23 This model automatically gets smarter as you collect more data. With enough data, it can predict which customers will churn with 85% accuracy. Once you have this insight at the individual customer level, you can confidently shift part of your marketing budget toward retention, which is where the real profit sits! ## The retention machine, where the real profit sits :) ### The simple math of value versus volume Let's use another simple example to show the difference between focusing on retention versus acquisition. Say an average webshop with โ‚ฌ1 million in revenue spends โ‚ฌ160K on acquisition and โ‚ฌ40K on retention. With that, they generate 2,000 new customers with an LTV of โ‚ฌ150. What happens if we shift to a 60/40 split (acquisition/retention)? The math: - โ‚ฌ120K acquisition โ†’ 1,500 new customers (25% fewer) - โ‚ฌ80K retention โ†’ retention rate rises from 20% to 35% - LTV rises to โ‚ฌ220 thanks to a better customer experience The result: - Old situation: 2,000 ร— 150 (LTV) = โ‚ฌ300K in value - New situation: 1,500 ร— 220 (LTV) = โ‚ฌ330K in value - Conclusion: 10% more value with 25% fewer customers ### How can you actually improve retention from a marketing perspective? You can probably come up with plenty of examples yourself by now, but to name a few good ones: - Work on personalization in your email channel, for example. The more relevant you stay, the more customers will feel that your company and offer fit them, and the sooner they'll come looking for you on their own. - Introduce a loyalty program. Tie it to predicted LTV, for example, and create tiers based on number of orders or order value (or a combination of both). - Set up proactive automated email flows based on lead scoring and churn prediction. For example, if you see that customers haven't been active or haven't purchased for a certain period, adjust your offers or your message accordingly. Every customer you keep active is almost pure profit! :) ## Conclusion Will you keep chasing volume while your CAC explodes? Or will you make the shift to value-driven growth? I'm phrasing it as a question, but the fact is you don't really have a choice in today's advertising arms race. You have to work on your retention (and therefore your LTV) to keep your CAC healthy in proportion. With every decision you make in your marketing activities, always ask yourself this question: "Does this increase customer value, or does it lower acquisition costs?" You can get started today by taking the first simple steps: 1. Calculate your LTV/CAC ratio (even with limited data) 2. Test the incrementality of your biggest marketing channel 3. Shift 20% of your budget from acquisition to retention These are big and bold decisions for many e-commerce companies, but in the end it's very clear to me: companies that focus on value survive, companies that chase volume disappear. Want to talk through your LTV/CAC optimization? I'm happy to help you think through the specific challenges in your situation. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### From LTV:CAC thinking to AI-driven growth: how e-commerce companies build a sustainable business model URL: https://www.martechnext.com/from-ltv-cac-thinking-to-ai-driven-growth-how-e-commerce-companies-build-a-sustainable-business-model/ Last updated: 2026-07-17T11:05:28.000Z This article is a deep dive following my earlier post on [LTV:CAC ratios](https://www.martechnext.com/focus-on-value-not-volume-why-e-commerce-companies-should-care-more-about-their-ltv-cac-ratio/). Where that piece made the case for value-driven marketing, this one gets into the concrete approach: how do you turn that mindset into working systems? Over the past years, I've spoken with dozens of e-commerce entrepreneurs who all ran into the same wall. They invest more and more in acquisition, watch their CPAs rise, and see their margins come under pressure. The solution they then choose? Pushing harder on those same channels. More budget into Google Ads, more variants to test on Meta, more discount campaigns to boost conversion. Not exactly sustainable. ## The problem sits deeper than your marketing mix In my previous article on LTV:CAC ratios, I described why e-commerce companies should focus more on customer value than on acquisition volume. But what does that mean in practice? And more importantly: how do you implement it when your team is already swamped with campaign optimizations and operational tasks? The core of the problem is that most e-commerce organizations have set up their marketing as a collection of separate silos. Paid search does its thing, the email team sends newsletters, the website shows the same homepage to every visitor, and CRM data sits somewhere in a system nobody really looks at. The result: every department optimizes for its own metrics, but nobody optimizes for the customer! ## The three pillars of sustainable e-commerce growth If you truly want to build a healthy LTV:CAC ratio, three fundamental shifts are needed in how you look at your business: ### 1\. From anonymous visitors to known profiles This is perhaps the biggest untapped potential for most webshops. On average, about 95% of your website visitors are anonymous. They browse, compare, and disappear again. You have no idea who they are, what they're looking for, or what would convince them. The traditional solution is retargeting: throw a pixel on the site and keep chasing them with ads. But that's getting harder and harder due to privacy restrictions, and on top of that, you're still treating everyone the same. The shift that's needed: from passively collecting data to actively building profiles. That means inviting visitors to tell you something about themselves in a natural way, without it feeling like filling out a form. Think of interactive elements that add value to their search, while simultaneously giving you insight into who they are and what they want. For exactly this, we developed an interactive preference wizard. Originally built for the travel industry, but the principle works for any e-commerce environment with choice overload: through a short, visual wizard, the visitor indicates their preferences. That input is instantly turned into a personalized product brochure and stored in your CDP (Customer Data Platform). Suddenly you no longer have anonymous sessions, but known profiles with concrete preferences. The effect is twofold: the visitor immediately gets more relevant content (which increases conversion), and you structurally build first-party data for all follow-up communication. ### 2\. From reactive marketing to predictive personalization Most e-commerce personalization I come across is reactive. Someone views a product, so you show related products. Someone abandons their cart, so you send an abandoned cart email. It's better than nothing, but it's always running behind the facts. The question you should be asking: can you predict what a customer wants before they know it themselves? This is where deep learning makes the difference. Not as a buzzword, but as a practical instrument. A well-trained model analyzes patterns in behavior: which products are viewed together, which profiles convert on which offers, which moment in the customer journey is the most crucial. The deep learning recommendation engine we built does exactly this. The system reads from your existing data warehouse (it doesn't store any data itself), analyzes behavioral patterns, and makes predictions along multiple axes: which products fit this visitor, which customer segment they belong to, and where in the funnel someone currently is. The crucial difference with traditional recommendation systems: deep learning sees patterns you would never discover yourself. And the system gets smarter with every interaction. ### 3\. From separate channels to an intelligent ecosystem The third pillar is perhaps the most underestimated: cross-channel intelligence. In most organizations, Google Ads learns nothing from what happens in the email channel. The website experience is disconnected from social retargeting. And customer service has no idea which marketing message someone just saw. The result: fragmented customer journeys, inconsistent messaging, and a lot of missed opportunities. What you want is a system where every touchpoint reinforces the next. Where the AI learns from email engagement to improve social targeting. Where website behavior translates directly into more relevant ads. Where customer service has context across the entire journey. This requires two things: a central data layer (BigQuery works well as a foundation, for example), and AI that can process and activate that data in real time across all channels. ## The practical implementation: from theory to working system I understand the above sounds ambitious if you're still wrestling with basic segmentation in your ESP (email platform), or maybe just getting started with your CDP. That's why the order of implementation is crucial. 1. **Start by centralizing your data.** Before you can deploy AI, your data needs to be accessible and reliable. That means: connecting all relevant data sources, creating a single customer view, and making sure your first-party data is centrally available. 2. **Begin with one high-impact application.** Not everything at once, but one concrete application that delivers results quickly. Often that's email personalization (relatively easy to implement, directly measurable effect) or the identification wizard (solves a concrete problem and immediately builds valuable data). 3. **Scale based on learnings.** Only when the first use case is running and paying off do you expand. Now you have data to justify further investments, and your team has built experience with the new way of working. ## The role of AI: automating intelligence Let me be honest about what AI does and doesn't do in this story. AI is not a magic button that solves all your problems. It's an instrument that enables you to do personalization at a scale that would be impossible manually. Where a marketer might previously have been able to come up with three customer segments and manually create content for them, AI can identify thousands of micro-segments and select the optimal content for each one. In real time, based on behavior, continuously learning. The approach that works combines those AI capabilities with human expertise. The recommendation engine predicts what customers want. AI agents turn those predictions into concrete personalizations. But the strategy, the brand positioning, the creative direction: that remains human work. ## What this means for your LTV:CAC Back to where we started: the LTV:CAC ratio as the basis for healthy growth. If you implement the three pillars above, you'll see shifts on both sides of that ratio. Your customer acquisition cost drops because your marketing budget is deployed more efficiently: you target more precisely, convert more, and reduce wasted ad spend. Your lifetime value rises because customers get more relevant experiences, spend more per order, and come back more often! The exact effect differs per company and depends on your current situation. But the pattern is the same everywhere: organizations that make the shift from volume-driven to value-driven marketing build fundamentally healthier businesses. ## The next step If this article matches your own situation, my advice is: start with an honest assessment of where you stand today. How well do you really know your visitors? How much of your marketing is personalized based on data versus assumptions? And how connected are your marketing channels, really? I regularly help e-commerce companies work through exactly this question, and then define a roadmap together that fits their ambitions and current maturity. No thick reports that gather dust, but concrete steps toward working AI. Curious what this could mean for your organization? I'd love to talk it through with you! \-- Bram Versteegh ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ### Birdsview raises โ‚ฌ2.5M seed to replace retention software and the agency that runs it URL: https://www.martechnext.com/birdsview-raises-eur2-5m-seed-to-replace-retention-software-and-the-agency-that-runs-it/ Last updated: 2026-07-17T09:59:45.000Z Here's a seed round that goes straight at one of my favorite topics: the gap between what retention software promises and what it actually does. Leipzig-based Birdsview has raised โ‚ฌ2.5 million to scale Avys, an AI agent that runs an online retailer's messaging channels end to end, writing an individual message to every customer instead of blasting segments. ## The details The round is led by Fortino Ventures (the firm behind Billy Grace, the fast-growing AI infrastructure platform for digital marketers in the Benelux) and Amsterdam-based Newion, an early backer of Parloa, now a billion-dollar AI agent company. TGFS backed the company at pre-seed. Birdsview was founded by data scientists Tim Kuesters and Ozan Brochwitz-Tuerker, who previously ran a CRM agency themselves. That biographical detail matters, because the entire pitch is built around making that business model obsolete. In the founder's words: "Avys isn't a better tool, it's the worker." The problem they're attacking is what Birdsview calls the double spend. Traditional email platforms give retailers the tooling, but someone still has to segment customers, build the flows, and write the content. So retailers hire an agency to operate the software, typically at three to five times the license cost. They pay twice: once for the tool, once for the people who make it useful. Birdsview's sharpest claim is structural: this problem persists by design. Agencies are the incumbents' biggest sales channel, roughly 40% of revenue by Birdsview's estimate. A vendor that automates the agency away dismantles its own distribution. Under the hood sits a product-intelligence layer the company built alongside its first hundred customers: 1.2 million labeled products across 60 industries, learning from 52 million email interactions per month. Instead of treating a product as a name and a price, Avys models what need it satisfies, which angle sells it, and when a customer is likely to need it again. Birdsview reports 3x more clicks and 8x more revenue per email for brands using Avys. ## The MartechNext take Regular readers know why this one caught my eye. The press release even quotes the same statistic I keep hammering on: customer acquisition costs are up 222% while retention, where the real profit sits, runs on software that hasn't fundamentally changed. Every euro of value you can squeeze out of your existing customer base improves your LTV:CAC ratio without spending a cent more on acquisition. A company attacking exactly that layer is, strategically, in the right place. I also genuinely like two things about the thesis. First, one-to-one instead of segments is the correct end state. Segmentation was always a workaround for the fact that humans can't write millions of individual messages. AI removes that constraint, so of course the segment eventually dies. Second, the channel-conflict argument is the most interesting sentence in the whole announcement. If your distribution depends on agencies, you will not build the product that replaces agencies. That's a classic innovator's dilemma, and it explains a lot about why the big ESPs ship AI features that assist the marketer rather than replace the workflow. Now the skepticism, because vendor-reported numbers deserve it. "8x more revenue per email" is a slippery metric: fewer, better-targeted emails will always look spectacular per email, and the customers who get personalized treatment are often already your most engaged profiles. I'd want to see incrementality, ideally from a holdout test, before celebrating. The same selection effect applies to every personalization case study, including the ones I've published myself. There's also an open question about control. Fully autonomous sending in your brand's voice is a trust exercise. One off-tone message to your best customers costs more than a quarter of agency fees. The winners in this category will be the ones that make the human-in-the-loop phase feel safe enough that retailers eventually dare to take the human out. And a note on the moat: every AI CRM startup claims its data compounds. Sometimes it's true. 1.2 million labeled products is a real asset, but the incumbents are sitting on vastly more interaction data. Their handicap isn't data, it's the distribution conflict described above. Birdsview's window exists exactly as long as Klaviyo and friends keep protecting their agency channel. That window is real, but it won't stay open forever. Verdict: small round, big thesis. This is one to watch, not because โ‚ฌ2.5 million changes the market, but because the "agent replaces tool plus agency" model is coming for every corner of martech. Retention email is simply where the math is most obviously broken. :) \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### From rising acquisition costs and data chaos to AI-driven growth in travel URL: https://www.martechnext.com/from-rising-acquisition-costs-and-data-chaos-to-ai-driven-growth-in-travel/ Last updated: 2026-07-17T09:19:51.000Z _This post is for subscribers only._ ### Meta Conversations 2026: the Business Agent is here, and it changes your WhatsApp strategy URL: https://www.martechnext.com/meta-conversations-2026-the-business-agent-is-here-and-it-changes-your-whatsapp-strategy/ Last updated: 2026-07-17T09:19:08.000Z On June 3, Meta presented its biggest WhatsApp announcement in years in London. Not everything is available in every market yet, but the direction is crystal clear. Here's what was announced, what you can do with it today, and why it confirms the case for conversational commerce, while raising an important strategic question about who owns your customer relationship. ## First things first: what is Conversations? Conversations is Meta's annual business messaging conference, and this year's edition was the fifth in a row. The event took place on Wednesday, June 3, 2026 in London, with a global livestream. On stage were, among others, Naomi Gleit (Head of Product at Meta) and Alice Newton-Rex (VP and Head of Product at WhatsApp). Why this conference matters: every year, it's the clearest signal of where Meta wants to go with business messaging across WhatsApp, Messenger, and Instagram. And the 2026 message is not subtle. For Meta, messaging stopped being a service channel a long time ago; it's becoming a growth, sales, and automation layer. The numbers underline it: Meta's AI tools now handle around 10 million business conversations per week, a tenfold increase since January 2026. ## The new features at a glance ### 1\. Meta Business Agent: the headline announcement The star of the conference is the Meta Business Agent: an AI agent that handles your customer conversations across WhatsApp, Messenger, and Instagram. You set it up in a few minutes, or connect it to your existing systems. What the agent can do: - Answer questions in your customer's language, in your brand's tone of voice - Recommend products from your catalog - Schedule appointments - Qualify leads - Close sales - Hand over to a human employee the moment the conversation calls for it This didn't come out of nowhere. Meta ran pilots for almost two years in markets like India, Mexico, and Brazil, and more than a million businesses were already using an earlier version on WhatsApp and Messenger. So this is not a cold launch, but the scaling up of something that was already running in practice. An important detail: getting started is free. But this is the first AI agent Meta explicitly plans to charge for. Paid subscriptions will roll out in the coming months along two tracks: smaller businesses pay through subscription tiers within WhatsApp Business Premium, while larger businesses pay based on token usage. The more the agent processes, the higher the bill. Meta has not yet published exact pricing. ### 2\. Meta Business Agent Platform: for larger organizations Alongside the ready-made agent, Meta launched the Business Agent Platform: an infrastructure layer that lets larger companies build, customize, and roll out agents at scale. The platform connects to hundreds of external systems, with Shopify and Zendesk among the named integrations, and offers enterprise controls, built-in safety frameworks, and measurability. ### 3\. WhatsApp Business Discovery: findability inside the app Meta is going to make it easier to find businesses within WhatsApp itself: by typing a business name into the search bar, or by sharing a phone number or contact card in a chat. Note: this feature was not yet live at the time of the conference. Meta called it an upcoming update. ### 4\. What's coming next Finally, Meta sketched a roadmap of features that aren't available yet, but show where things are heading: - Daily briefings that summarize the night's conversations and surface insights (now in limited testing on WhatsApp Business, Instagram Pro, Messenger, and Meta Business Suite) - Market research and surfacing product insights - Calendar management - Competitive intelligence The common thread: from giving individual answers to structuring the work around them. Think triage, prioritization, and the logical next step. ## What's actually available in your market, and what isn't? This is where you need to stay sharp. "Global rollout" and "available in your market as promised" are not the same thing. A sober look at the current state: **What's live.** The Business Agent has launched as a global product and is, at its core, a Meta-hosted service. EU businesses can in principle use that hosted agent, and getting started is free. WhatsApp Status Ads and Promoted Channels, which I covered earlier, have been rolling out across European markets; in the Netherlands, for example, they've been live since late February 2026\. WhatsApp campaigns can be managed through the same Ads Manager as Facebook and Instagram. **What's not settled yet.** The paid subscriptions (WhatsApp Business Premium tiers and token pricing) will only roll out in the coming months. WhatsApp Business Discovery isn't live anywhere. The briefings and the entire "coming soon" list are in limited testing, without a concrete EU date. **The EU caveat you can't ignore.** Meta has a consistent pattern in Europe: AI features arrive here later, and often stripped down. Meta AI itself took about eight months longer to reach the EU than the US, and started with a slimmed-down feature set because of the strict privacy regime. There's something else at play too: since June 3, EU businesses are not allowed to self-host Meta's Llama 4 models under the license. That's not a blocker for the hosted Business Agent, but it does illustrate Meta's room for maneuver in Europe. And the model underneath the Business Agent? Meta hasn't disclosed it. In short: the direction is right, but don't count your chickens. Check per feature what is actually switched on for your account in your market before you build your strategy on it. ## The real question: rent from Meta, or own your intelligence layer? Let's be honest: this announcement is good news for everyone betting on conversational commerce. With the Business Agent, Meta is building out exactly the thesis that independent conversational commerce stacks have been built on: WhatsApp is not an extra newsletter, it's the channel where acquisition, service, and sales come together. When the biggest player in the world validates your thesis, you're in a good place. But there is an important difference between renting from Meta and keeping control yourself. And that's where the strategic choice sits. The Business Agent is a generic agent, inside Meta's walls, on Meta's terms, running on an undisclosed model, and soon billed per token or per subscription. The more it works, the more you pay. That's the classic walled-garden mechanism: convenient to start with, expensive and dependency-creating to scale on. The alternative is an intelligence layer you own, spanning your whole funnel instead of sitting on top of one channel: - **Acquisition**: personal 1-on-1 campaigns via WhatsApp, with AI replies and predictive product recommendations powered by a recommendation engine. No bulk, no broadcast: genuinely personal contact, with attribution all the way down to revenue. - **Guidance**: a whitelabel WhatsApp assistant that supports your customer 24/7, during and after the purchase, under your own brand and connected to your own backend. When things get uncertain or complex, it hands over warmly to a human. Together, they cover the entire customer journey: from campaign to conversation to fan. Where most setups stop at the click, this approach makes sure the click converts. So the question you should be asking yourself as an advertiser is not "should I use the Business Agent?". You can absolutely deploy it where it fits. The real question is: do I want to place my customer relationship and my data inside Meta's environment, or do I want to own that capability as a layer over my own channels? The more dependent and expensive the platforms become, the more valuable your own intelligence layer and a solid first-party data foundation become. ## What you can do now A few next steps: 1. **Stop waiting on WhatsApp.** With 3 billion users worldwide and 90%+ open rates, it's a serious growth channel by now, not an experiment. 2. **Activate the Business Agent**, but factor in the token and subscription costs when scaling, and verify what's actually live in your market. 3. **Build your own layer.** Combine acquisition, journey-long guidance, and prediction in a stack you control, so you're not locked into one platform and one pricing model. With Conversations 2026, Meta has set the tone: the future of marketing is a personal conversation. I fully agree. The only question is on whose terms you'll be having those conversations: the platform's, or your own. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime. ### Zoey launches: WhatsApp journey automation for travel, and yes, this one is mine URL: https://www.martechnext.com/zoey-launches-whatsapp-journey-automation-for-travel-and-yes-this-one-is-mine/ Last updated: 2026-07-18T18:26:38.000Z Let me start with the disclosure, because you deserve it before anything else: Zoey is my product. I'm the founder. So no, this is not independent coverage, and I won't pretend it is. What I can offer instead is something you rarely get in a launch announcement: the honest reasoning behind why this product exists, from the person who has been writing about the underlying problem on this platform for months. ## What Zoey is Zoey is a WhatsApp journey automation platform for the travel industry, built for tour operators, OTAs, holiday parks, and travel agencies. It guides travelers automatically and personally from booking to homecoming and beyond, via the one channel virtually every traveler already uses daily. The platform consists of four building blocks: - A visual Journey Builder for proactive tip timelines, so guests get the right suggestion at the right moment of their trip - An AI engine that automatically answers 80% of customer questions - A WhatsApp newsletter module with open rates above 90% - A real-time dashboard with analytics and a full conversation overview It works for very different travel concepts: a tour operator with complex long-haul itineraries has other needs than a holiday park that wants to inspire guests locally, or an OTA that wants to personalize at scale. Zoey is live now for travel companies in the Netherlands and Belgium, and you can request a demo at withzoey.com. ## Why I built it Regular readers will recognize the thesis immediately, because it runs through almost everything I publish here. The travel industry has a retention problem. Acquisition costs keep rising, yet the vast majority of travelers book their next trip elsewhere, simply because contact drops off the moment they get home. The relationship, and with it the next booking, quietly dies in an inbox with a sub-20% open rate. Meanwhile, the moment of highest engagement in the entire customer journey, the trip itself, is exactly where most travel companies are absent. Nobody is there when it rains in Barcelona and the family needs a plan B. Nobody suggests the museum tickets, the beach club, the excursion that fits this specific travel party. That's the gap Zoey fills. During the trip, travelers can book admission tickets, excursions, and activities directly via WhatsApp, which gives travel companies margin beyond the initial booking. After the trip, the platform builds on the profile it developed (travel behavior, preferences, conversations) and automatically generates personalized recommendations for a next destination. The conversation simply never stops, and that's the whole point: retention is not a campaign, it's a relationship that stays warm. If you've read my piece on intelligent personalization or the travel growth roadmap, you've seen the building blocks of this thinking before. Zoey is what happens when you stop writing about the flywheel and start shipping it. :) ## The honest caveats Because I hold vendor claims to a standard on this platform, mine don't get a pass. The 90%+ open rate is a channel statistic: WhatsApp messages simply get opened, which is precisely why the channel is valuable, but opens are not bookings. The "80% of questions answered automatically" number is real but will vary by travel concept and how well your data foundation is set up. And as I wrote in my WhatsApp marketing piece: this channel punishes overuse. Frequency discipline and genuine relevance are not optional, they're the difference between a trusted travel companion and spam. I'll also say this: I will keep covering competitors and adjacent players in this space the way I cover everyone, and when I write about the conversational commerce category, I'll remind you of this disclosure. If I ever fail at that, call me out. My credibility here is worth more to me than any single product plug. ## Try it If you run a travel company in the Netherlands or Belgium and the retention story above sounds painfully familiar, take a look at withzoey.com and request a demo. And if you're a reader from another industry: the pattern (own the conversation, build the profile, never let the relationship go cold) translates far beyond travel. That's a topic for another article. \-- Bram Versteegh --- *Bram Versteegh is the founder of MartechNext, covering the business of AI in marketing: who's building it, who's funding it, and how industries put it to work. He is also the founder of Zoey, the platform discussed in this article.* **Enjoyed this piece?** Every week, What's Next delivers sharp analysis, the latest funding rounds in AI marketing, and the martech news that actually matters. Straight to your inbox, free. ## Get What's Next in your inbox Every week I break down what's actually happening in AI marketing: the funding, the tools, and what it means for your business. Count me in! Email sent! Check your inbox to complete your signup. Join marketers from travel, e-commerce and beyond. Unsubscribe anytime.