AI Shopping is already happening. And I’ve got the agency data to prove it.
I talk to a lot of people about this lately. Clients. People I meet at events. Even our own SEO specialists and account managers. And I keep hearing the same two patterns.
Pattern one: “Yeah, we need to do something with it. After the summer. Or end of the year.”
Pattern two: “Nah, we looked at our GA4 numbers, it’s not worth it yet.”
And then there’s a third pattern. My reaction. Which is, every single time, an internal wtf.
Here’s why.
Since the end of 2025, in the Netherlands and the countries around us, we started seeing Product Boxes in ChatGPT. It went visual, actual product tiles instead of plain citation links. If you don’t see them, you might be on a paid account. Open an incognito window and check. Since I noticed that, something in my head switched on. I got curious.
For me it’s crystal clear. AI Shopping is the biggest trend out there right now. And almost everyone is acting like it’s barely a thing.
Anyone who wants to argue the opposite: come challenge me. I’m backed by our agency data across 150+ Google Analytics properties of ecommerce clients.
First, some facts from people you’ll actually believe
I’m usually the guy saying don’t believe everything you read on LinkedIn or on the internet. But in this case, these might help move you. And then I’ll show you my own data, because I always want to see it for myself.
- 28% of Gen Z say they already use gen AI tools for shopping (vs 16% of baby boomers), in McKinsey’s survey fielded 31 March–8 April 2026.
- Adobe’s tracking shows AI referrals leading on conversion: revenue per visit from AI sources up 254%, and AI shoppers 33% less likely to bounce than other traffic.
- 64% of consumers plan to use AI chatbots for shopping in 2026, and nearly 1 in 4 plan to make AI shopping their default way to shop.
- ChatGPT grew from under 1% of Amazon’s weekly search volume to over 8% in less than a year, and now takes over 84 million shopping questions from U.S. consumers every week.
- ChatGPT processes an estimated 50 million shopping-related queries per day, against 800–900 million weekly active users.
- For the first time in history, annual electronics search volume across Amazon, Walmart, Target and Best Buy fell over 5% year-over-year, as shoppers moved discovery to places like ChatGPT.
- 61% of consumers have used GenAI tools like ChatGPT for online shopping.
- McKinsey projects the agentic shopping channel will drive $3–5 trillion globally by 2030; Morgan Stanley’s AlphaWise survey shows LLM adoption approaching 50% in the US, with AI agents capturing 10–20% of e-commerce ($190–385 billion).
Impressive numbers. But I don’t run on other people’s dashboards. So I did the research myself, at least for the clients we work with.
And no, I’m not a data scientist. I’m just a curious guy with an MCP, a giant pile of data, and a genuine obsession with where this is going.
The research and how we did it
Methodology (short version):
- Only ecommerce clients
- Across European countries (comparable ChatGPT behaviour, since EU law shapes how it operates here)
- Period: July 2025 – June 2026
And attribution is hard, I know it. GA4 last-click understates assisted AI journeys. The product-page classification is directional, not audited to the exact percentage point. And the product-page views are based on a client subset, so read them as shape and dominance, not precise portfolio totals.
Here’s a little look inside how my brain works when I dig into this.
My question: “Are people arriving from ChatGPT actually landing on product pages, and is that growing?”
What you’re seeing: product-page sessions from ChatGPT, per month (left) and per quarter (right). Both climb steeply. Black Friday (Nov 2025) is marked as the old high point. And here’s the standout: May and June 2026 are the two highest months ever recorded, both above Black Friday, and Q2 2026 is the biggest quarter yet. So ChatGPT visitors landing straight on a product page aren’t just growing, they’ve hit record levels in the most recent months, beyond the usual holiday peak.
What about revenue?
My question: “The visits are fun. But what about revenue? Is that breaking records too?”
Total ChatGPT-attributed revenue across all clients, per month and per quarter. Same record-breaking story, and it’s even stronger: May and June 2026 each brought in more ChatGPT revenue than Black Friday, the two biggest revenue months of the year, and Q2 2026 is the highest-revenue quarter. The visits are turning into real money, at record levels.
What about the last couple of months?
My question: “What about the last couple of months, how much of total revenue is ChatGPT bringing in now?”
I’ll be honest, this one’s a strange visual: percentages on both the X and the Y axis. Let me explain it properly in the next one.
What you’re seeing: how much of each client’s full-year ChatGPT revenue landed in just the last quarter (Apr–Jun 2026). 73% of clients earned more than a third of their entire year’s ChatGPT revenue in those three months. About half (49%) earned more than 40%. And 10% of clients earned more than 60% of their whole year in that single quarter. The takeaway: this is a very recent surge. Most of the year’s ChatGPT revenue is landing right now, not spread evenly across the year.
Which leaves me with the question that should scare you: what if?
The what if. What if Google flips AI Mode to default? And don’t you dare tell me they won’t, they already rolled out conversational shopping attributes in Merchant Center. The plumbing is being laid right now, in the open. In some countries, like mine, there are no products in AI Mode yet. Give it a few months.
And when that switch flips, what exactly is your plan? Hope people still scroll to ten blue links? They won’t. What’s left is people jumping to the product tab, to marketplaces, window-shopping on images and prices. But look at the actual data: people are already getting used to buying after a conversation. Soon they won’t need to leave the chat at all. The doorway you’ve optimized for a decade is quietly being bricked up, and most of you are still arguing about whether to look at it after summer.
What should you do? Optimize Google Shopping.
Huh? That old-school thing? I thought this was all about ChatGPT?
It is. But here’s the thing, we actually know how ChatGPT works: you prompt it (“what’s the best X for Y?”), it runs a query fanout, reads an independent review because it won’t make the call on its own, then does another fanout for those products in Google’s organic Shopping tab, judges what sits in the first 40 results, and pulls the winners straight back into the answer.
That’s a long sentence, I know. But that’s the mechanism. And to be in that answer, you need to be in organic Shopping.
The math nobody does
This might be the most important paragraph of the whole piece. For twenty years we’ve been spoiled rotten. A click was a click, a click became a session, a session became revenue, and the dashboard tied a neat little bow on all of it. Easy. Comfortable. Measurable to the decimal.
That era is over. Attribution is broken now, and the click is no longer the story. If you keep staring at the same GA4 report waiting for it to tell you AI is happening, you’ll keep concluding it isn’t, because the report literally can’t see most of it. So look behind the click.
Besides analytics, we also track AI visibility, both for us and for our merchants (our clients), across a lot of prompts. So we know, roughly, on how many commercial prompts there’s at least one shopping box or carousel. And we know our answer share.
Here’s the thing. For a lot of our clients, we only appear in 5% to 20% of the answers where there’s a product box. So if there’s a product in 100 answers, we’re in 5 to 20 of them.
Now do the math. If your potential ceiling is 100% and you cover 5%, that’s 100/5 = 20x headroom. Then: barely anyone clicks through today, well under 1%, so multiply again. And you’re usually sharing that answer with about 3 other merchants, so divide. This isn’t a clean revenue projection. I’m hammering it into you on purpose: this is how much AI search is already being used to find products, and you cannot afford to ignore it.
Make it concrete. Say ChatGPT sends you €1,000 in revenue this month, and you only show up in 5% of the relevant answers. Lift that to full coverage and you’re at roughly €20,000 (20x). Now remember barely 1% clicks through today, so the actual buying intent flowing past you is orders of magnitude bigger again. Even split across a handful of merchants, that fictional €1,000 represents a demand pool in the hundreds of thousands. That’s what “not worth it yet” actually looks like.
And with shopping boxes arriving, we get a little of our good old clicks back, because people don’t need to leave the platform anymore. They can click when they’re actually ready to buy.
So please, stop putting this on the long-term pile. The shift is exponential. Every day you ignore it, you’re ignoring revenue that could be yours.
Oh, and one last thing. I don’t think ChatGPT is going to be the one that wins. If Google wanted to, they could kill it. But that changes nothing about the advice above, it just means Google (AI Mode) replaces ChatGPT as the front door. Same game, different logo.
That’s for another article.
