“Our AI Traffic Is Up.” “Did Any Of Them Buy?”
Last month’s review meeting went fine right up to the traffic chart. AI referrals were up again, month over month.
Then somebody asked the obvious follow-up: did any of those people buy anything?
I said I’d pull the numbers and get back to them. So, probably, has everyone else who’s sat through that exact meeting.
But the cell where that answer belongs is still blank. Measuring whether AI sends someone your way is one question. What they do once they land is a different one, and most of us aren’t tracking it.
A new consumer survey went looking for that answer — what an AI-referred shopper actually does next.
Does AI Beat Influencers On Trust?
For an unfamiliar brand, yes — 67% of shoppers say they’d trust an AI’s recommendation over a social media creator or influencer’s.
Contentsquare put that question, and others, to 2,000 shoppers across the US and France, publishing the results on its own site on 2026-09-03 (R).
Contentsquare sells site-experience analytics. The same page that reports these numbers ends on a pitch for its own platform — a stake in the “your website matters” conclusion, worth keeping in mind.
Having a stake doesn’t make a number fake. It’s a reason to read the rest of what they measured before deciding what to do with it.
AI Already Moves Real Spending
47% say AI has changed their mind about which brand to buy. Not whether to buy — which one.
57% went further: AI recommendations have already pulled real money out of their pocket, not just their opinion.
Shopping habits back that up. 36% say they now use AI instead of a search engine for at least some of their shopping — a limited claim, some of their shopping rather than all of it, but not a rounding error either.
By age, adoption skews toward the middle: 74% of 30-44-year-olds say they’re likely to use AI while shopping online, against 70% of 18-29-year-olds and 65% of 45-60-year-olds. Every group clears 65%.
If “our customers aren’t there yet” is still the working assumption on your team, this is the number that assumption has to clear.
What Happens Next Is Still Yours To Shape
Here’s the number I’d actually put in a deck: 23%.
Contentsquare asked what shoppers would do if the site an AI sent them to turned out to be disappointing. Only 3% said they’d buy anyway. The other 97% said they’d reconsider their next move.
That 97% is a hypothetical answer — “would,” not “did.” It isn’t a measured abandonment rate, and it shouldn’t travel as one.
The 97% splits three ways: 57% would keep researching, 23% would switch to a different brand, and 16% would abandon the purchase altogether.
The 57% still researching hasn’t left. The 16% who’d abandon the purchase weren’t headed to a competitor either — that’s lost demand, not lost share. The one group that’s an actual loss to a rival is the 23% who’d switch brands.
All three numbers answer the same conditional: what a shopper says they’d do if your page disappoints an AI-referred visitor. Change what’s on the page, and you’re working the one variable this survey says is still yours.
The Gap Sits Between What AI Says And Your Own Page
21% say an AI shopping assistant gave them information that didn’t match what they found on the brand’s website. Another 20% said the AI didn’t give them enough detail to buy with confidence.
Neither gap shows up from staring at your own site. You’d have to go look at what the AI is actually saying about you — the same problem ChatGPT Cited Your Brand. Now Go Read What It Said. covers from the citation side.
A few limits are worth holding onto before this number travels anywhere else. Contentsquare surveyed shoppers through Pollfish, an online polling platform — this is what people say they’d do, not a log of what they actually did. The sample is the US and France; there’s no Japan data here. The 2,000-person total is weighted and stratified, but Contentsquare doesn’t publish the fieldwork dates, so there’s no way to say which month of AI behavior these answers describe. And Contentsquare, as above, sells the kind of platform this finding argues for.
None of that makes the 67% or the 23% wrong. It means they’re a snapshot from an interested party, not a number you can defend on its own.
Put 23% In The Deck, Then Check What The AI Is Saying
97% and 23% come from the same question, asked of the same shoppers. I’d only carry 23% into a report — the one that names an actual shift to a competitor, with the “if the site disappoints” condition attached.
The other move costs less than it sounds like: once a month, pull what an AI is telling shoppers about your brand and set it next to your own page. The 21% gap between the two doesn’t show up any other way.
The quality of who AI sends your way is its own question — Turns Out AI Traffic Beats Paid Social, Especially On Complex Products has the numbers on that side. This one is about what happens once they land.
Does your reporting have a line for what AI is actually saying about you? If not, that’s the gap worth closing before the next AI-referral number goes into a slide.
Sources
- Contentsquare, “97% of Shoppers Rethink an AI-Recommended Purchase When the Website Falls Short, Contentsquare Finds”, 2026-09-03, https://contentsquare.com/press/ai-recommended-purchases/
