Imagine a patient sitting in the chair. The dentist says, “You need a crown on that molar.” The patient nods, thinks for a second, and says, “I’d like a second opinion.” The dentist says, “Of course.” The patient turns back to the same dentist and asks, “So… do I need a crown?”
That’s absurd. Nobody does that.
Except that is exactly what your buyers are doing right now — and a study caught them on camera.
Hi, I’m Jeff Payne. You’re listening to Episode 93: Only 3 in 100 Looked!
Let’s start with a number everybody got comfortable with. A survey this year found that only 11% of consumers are willing to let AI make a purchase for them. 11%. And a lot of business owners heard that and relaxed. “See? People still want to decide for themselves.”
They do want to decide. They don’t want to shop.
Here’s what I mean. A behavioral study observed 56 people as they worked through 221 real buying decisions within an AI assistant — screen recordings, think-alouds, the whole thing. Not what people say they do. What they actually did.
93% of those sessions ended with no click to the open web. None. The AI built a shortlist, the buyer read it, and the buyer chose. Your website never loaded.
And here’s the one that gave this episode its name: three percent verified anything on an outside site. Three in a hundred looked. Nobody — not one person — cross-checked a single price.
Now, compare that to what people tell survey takers. In a separate survey, 54% said they double-check everything AI tells them. 54% say they check. Three percent actually did. That’s the gap between what your buyer believes about themselves and what they do at nine o’clock at night with a phone in their hand.
So where did the second opinion go? That’s the part that stopped me cold.
Verification didn’t disappear. It moved inside the chat. In almost 40% of those sessions, the buyer did push back — by asking the AI to justify its own answer. “Why that one?” “Are you sure?” And the AI explained itself, and the buyer said, “Great,” and moved on.
That’s the patient asking the same dentist. The judge checked the judge.
And it worked — for the AI. Buyer confidence went up in 82% of sessions. It went down in none. Zero. People walked away more certain of their choice no matter which list they were handed. Read that again, because it’s the whole episode: your customer’s confidence is no longer about you. It’s about the list.
A second study, this one inside Google’s AI Mode, found the same shape. When people used classic search, more than half built their own shortlist — they compared, they rejected, they discovered. In AI Mode, five percent did. 88% took the list as given. And three out of four picked the top name on it. Third place and below? 10%, combined.
So the funnel didn’t get shorter. A stage got removed. The stage where the buyer compares you to the alternatives — the stage your website was built for — happens now before they ever arrive, in a room you’re not in, from evidence you didn’t submit.
What evidence is that? An analysis of 75,000 brands asked which signals align with being named in AI answers. The strongest, across every assistant: being mentioned on YouTube. Then being mentioned on other people’s websites. Then other people linking to you by name. Your domain rating — the number so many of us chased for a decade — correlates at less than half the strength of a YouTube mention. The size of your site barely registers.
Think about what that list has in common. Every item on it is something somebody else says about you.
There is exactly one override in all of this data. When the buyer already knows a name, recognition beats rank one. In the AI Mode study, shoppers leaned on brand recognition as their trust signal in a third of sessions — half again as often as they did in classic search. The only buyer who doesn’t sign the list is the one who walked in already knowing who they wanted.
That’s the Compounding Cue arriving at its destination. Every mention, every review, every time your name gets said on a channel you don’t own — that’s the deposit. This is the withdrawal.
So here’s the move, and it’s a reallocation, not a new tactic.
Stop over-investing in the room the buyer never enters. Your website still has to be right — remember, whatever the AI says about you stands as fact, because nobody lands on your site to be corrected. But “right” is the floor now. It is not the strategy.
The strategy is producing the exhibits the model weighs. Answer the questions your buyers are actually asking AI — not the questions you wish they’d ask — and answer them on the surface AI cites most, which is video. Give other people a reason to write your name. Earn the review, the mention, the link that carries your name.
I call this building Highlight Reels: short, professional, question-by-question answers, made to be found and cited. It’s hard work. That’s the point. Most businesses are still cranking out posts that nobody reads, and that cite no models. The hard thing is the moat.
Three diagnostic questions before you go.
#1. Open an AI assistant and ask it for the best option in your category and market. Are you on the list? Where? And read the downside it attaches to you — because one disclosed negative was enough to knock a brand off the list in that study.
#2. Ask it about you by name. Does it know who you are, or does it guess?
And #3. Of everything you’ll spend on marketing this quarter, how much of it produces something somebody else says about you?
Because your buyer isn’t shopping anymore. They’re signing. The only question is whose name is on the list.
I’m Jeff Payne. Thanks for listening. I’ll see you next time.
Imagine a patient in the chair. The dentist says, “You need a crown.” The patient asks for a second opinion — then returns to the same dentist and asks, “So, do I need a crown?”
Nobody does that. Except, as a behavioral study has now documented on camera, that is exactly what buyers do when they shop with AI. I want to talk about the quietest change in buying behavior that I have seen in 25 years of marketing: the buyer stopped shopping and started signing.
They won’t let AI buy. They let it choose.
A Gartner survey this year found that only 11% of U.S. consumers are willing to let AI make a purchase decision for them. Plenty of business owners read that number and relaxed: people still want to decide for themselves.
They do. They just don’t want to shop. A joint behavioral study by Profound, Kevin Indig, and Clickstream Solutions watched 56 people work through 221 real buying tasks inside ChatGPT — screen recordings, think-aloud narration, post-task surveys. Not what people say they do. What they did.
93% of sessions ended with no meaningful click to the open web. The AI built a shortlist;tlist, the buyer read it; the buyer chose. Only 3.2% verified anything outside the site; a single participant cross-checked a single price externally.
Compare that to self-report. A separate Gartner survey found 54% of AI shoppers say they double-check everything AI tells them. 54% say they check, but only 3% actually did. That gap — between what buyers believe about themselves and what they do at nine o’clock at night with a phone in hand — is the whole story.
54% say they check. Only 3% actually did.
The judge checked the judge
Verification didn’t disappear. It moved inside the chat. In 38.8% of sessions, the buyer pushed back — by asking ChatGPT to justify its own answer. The model explained itself, the buyer said “great,” and moved on. That’s the patient asking the same dentist.
And it worked, for the AI. Buyer confidence rose in 81.9% of tasks and fell in none. People walked away more certain of their choice regardless of which list they were handed. Your customer’s confidence is no longer about you. It’s about the list.
A second study, Growth Memo’s AI Mode usability research, found the same shape inside Google. In classic search, 56% of participants built their own shortlist — compared, rejected, discovered. In AI Mode, 5% did. 88% accepted the list as given, 74% picked the top-ranked option, and everything from rank downward accounted for about 10%.
Your customer’s confidence is no longer about you. It’s about the list.
A stage got removed, not shortened
The funnel didn’t get shorter. The stage where a buyer compares you to the alternatives — the stage your website was built for — now happens before they arrive, in a room you aren’t in, from evidence you didn’t submit.
What evidence? Ahrefs analyzed 75,000 brands to see which signals line up with being named in AI answers. The strongest across ChatGPT, AI Mode, and AI Overviews: YouTube mentions, at roughly 0.74. Then branded web mentions (0.66–0.71). Then branded anchors — other people linking to you by name. Domain rating sits at 0.27–0.33, less than half the strength of a YouTube mention, and the number of pages on a site barely registers.
Look at what the top of that list has in common. Every item is something somebody else says about you.
This is one of the best written, most informative email newsletters I’ve ever read!
Thank you for sharing your expertise, and hope you’re doing well.
Beth Zimmerman
Founder, Executive Director
The one override: a name they already know
There is exactly one exception in all of this data. When a buyer already recognizes a brand, recognition beats rank #1.
In the AI Mode study, shoppers cited brand recognition as their trust signal in 34% of sessions, compared with 23% in classic search. The only buyer who doesn’t sign the list is the one who walked in already knowing who they wanted.
That is the Compounding Cue arriving at its destination. Every mention, every review, every time your name gets said on a channel you don’t own is the deposit. The AI shortlist is the withdrawal.
Recognition beats rank #1. It’s the only thing that does.
The Move: fund the exhibits, not the room
This is a reallocation, not a new tactic. Stop over-investing in the room the buyer never enters. Your website still has to be right — whatever the AI says about you stands as fact, because nobody lands on your site to be corrected. But “right” is the floor now, not the strategy.
The strategy is producing the exhibits the model weighs. Answer the questions your buyers actually ask AI, on the surface AI cites most, which is video. Give other people a reason to say your name in writing. Earn the review, the mention, and the link that carries your name in it.
I call this building Highlight Reels: short, professional, question-by-question answers, made to be found and made to be cited. It’s hard work. That’s the point. Most businesses are still cranking out posts nobody reads and no model cites. The hard thing is the moat.
The hard thing is the moat.
ThREE DIAGNOSTIC QUESTIONS
One. Ask an AI assistant for the best option in your category and in your market. Are you on the list? Where? Read the downside it attaches to you — in the ChatGPT study, one disclosed negative was enough to knock a brand off a shortlist.
Two. Ask it about you by name. Does it know who you are, or does it guess?
Three. Of everything you’ll spend on marketing this quarter, how much produces something somebody else says about you?
Your buyer isn’t shopping anymore. They’re signing. The only question is whose name is on the list.
Source: Kevin Indig, Growth Memo — “Shopping is part of the Product” (the prompt for this episode) and “How consumers navigate high-stakes purchases in AI Mode” (AI Mode usability study: AI Mode n=56, classic search n=36) — https://www.growth-memo.com/p/how-consumers-navigate-high-stakes
Profound, Kevin Indig and Eric Van Buskirk (Clickstream Solutions) — “The shortlist is the new shelf,” July 2026 (56 participants, 221 codable ChatGPT shopping tasks) — https://www.tryprofound.com/blog/the-shortlist-is-the-new-shelf
Ahrefs, Louise Linehan and Xibeijia Guan — “Top Brand Visibility Factors in ChatGPT, AI Mode, and AI Overviews (75k Brands Studied)” — https://ahrefs.com/blog/ai-brand-visibility-correlations/
Gartner — “Gartner Survey Finds Consumers Want AI Shopping Help, But Not AI Purchase Decisions,” May 27, 2026 (322 U.S. consumers, January 2026; 846 U.S. consumers, Nov–Dec 2025) — https://www.gartner.com/en/newsroom/press-releases/2026-05-27-gartner-survey-finds-consumers-want-ai-shopping-help-but-not-ai-purchase-decisions
COMPLETE THE FORM TO
BOOK A STRATEGY CALL
"*" indicates required fields
COMPLETE THE FORM TO
BOOK A STRATEGY CALL
"*" indicates required fields
Subscribe and Share – WE APPRECIATE YOUR SUPPORT
