The other night I was about to buy an egg peeler.
I’m serious. I was on the checkout page. Bundle picked, payment buttons sitting right there. Forty seconds from owning two egg peelers!
And then I did something I’ve started doing without really thinking about it — I asked Gemini whether the egg peeler was legitimate.
The results from Gemini came back in a few seconds. The interesting part was that it wasn’t the verdict, and the results never left the page.
It didn’t go research the company, or the product, or the category. It read what was already sitting there, in public, on the seller’s own website — and it found things on that page that couldn’t all be true at the same time.
I closed the tab to the egg peelers website.
Hi, I’m Jeff Payne. You’re listening to The Jeff Payne Show, Episode 84: The Page Testified Against Itself.
Last time we talked about how a claim becomes a fact — how something untrue gets repeated enough times that nobody remembers to check it. That was the supply side. Somebody manufacturing belief.
This is the other half. The demand side. What happens when the buyer installs a checking step of his own.
Same technology, opposite job. In that story, the AI was the thing that failed to verify. Here, it’s the thing doing the verifying. The only difference is that somebody asked.
And more people are asking every week. Not the way we used to search, looking for options. The way you’d hand a contract to a skeptical friend before you sign it.
Here’s what actually convinced that page, and it matters more than the product ever could.
The photograph at the top showed a raw egg being cracked open. The product underneath it was sold as a peeler for already-cooked eggs. Two different jobs.
Then the reviews. One praised the hand-cranked design. The one directly beneath it described dropping the egg in a container and giving it a few good shakes. Same product. Two mechanisms that can’t both exist.
And a customer count in the six figures for a gadget nobody has ever heard of.
Now — here’s the part I keep coming back to. That page was covered in trust signals.
Five stars.
A security badge.
A guaranteed-safe-checkout seal.
A 90-day money-back guarantee.
Every major payment logo lined up in a row.
Every single credibility marker a buyer has ever been taught to look for.
All present.
All passing.
None of them mattered.
Because badges answer a question nobody was asking anymore. They tell you somebody went and got a badge. They don’t tell you whether the story holds together. And nobody had to know a thing about eggs to catch that it didn’t.
No research.
No outside source.
No expertise of any kind.
The page was cross-examined against itself, and it lost.
That’s the part I want you to sit with.
For years, the assumed risk has been what other people say about you. The review you can’t take down. The article that got it wrong. That risk is real, and we’ve covered it.
This is a different exposure, and it’s entirely your own. It’s the things you published yourself at different times for different reasons that no longer agree.
The About page written in 2019.
The press release from the acquisition.
The service list on your website that doesn’t match the service list on your Google Business Profile.
The bio page with the old title and outdated content.
The case study with a number that contradicts the number in the other case study.
Every one of those was harmless yesterday — because nobody read them together. A human buyer skims one page. A human buyer never opens 11 of your properties side by side and compares the claims line by line.
That’s now a 4-second job, and it’s free. And it happens at the exact moment somebody is deciding whether to call you.
Drift used to be invisible. Now it reads as a verdict.
And notice what the verdict actually is. It isn’t, “This company is wrong!” It’s something worse and stickier: This company can’t be trusted to describe itself accurately.
A buyer will forgive being wrong about something. Nobody, however, hires somebody they’ve decided is unreliable.
Here’s the receipt on this, and it’s the part that should make you uncomfortable.
I wasn’t researching.
I wasn’t comparing options.
I had made the decision.
The bundle was selected.
The payment buttons were on the screen.
And what stopped that sale wasn’t a bad review somewhere else. It wasn’t a warning from a friend, or a scam database, or an article somebody wrote. Nothing outside that page participated at all.
The seller’s own website did it.
Their photograph.
Their reviews.
Their numbers.
Every piece of evidence used against them, they had published themselves, on purpose, to sell me something.
In less than a minute, when I was ready to click the buy button on a decision already made.
So here’s the shift, and it’s a mindset, not a project.
Stop thinking about your web presence as a set of marketing assets. Start thinking about it as testimony. Because that’s how it’s being read now — not one page at a time by a distracted human, but all at once by an AI Agent, such as Gemini, specifically looking for the places, in nanoseconds, where your story doesn’t line up.
The old standard was: Is each marketing asset good, or good enough? Is it credentialed, is it badged, is it certified?
The new standard is: DO THEY ALL AGREE?
Those are different tests, and the new standard is harder — because almost nobody in your company owns it. Marketing owns the website. Somebody else owns the Google Business Profile. You own your own LinkedIn. The press release is four years old, and nobody has read it since the day it went out. Nothing is wrong, but nothing agrees.
Your awards page, your years-in-business line, your certifications — all of that is credentialing. And credentialing is now the easy test. It was always the easy test. It just used to be the only one anybody ran.
The question worth sitting with is an uncomfortable one. If somebody pointed a machine at everything your company has ever said about itself and asked one question — “Does this all hold together?” — what results would an AI agent, such as Gemini, find?
Not, “Would it find something good?” Would it be in agreement?
And you don’t have to guess. Somebody is already running that check on you. They’re just not telling you the result.
Proof over proximity — but proof only counts when it corroborates. Two of your claims that contradict each other don’t cancel out to neutral. They read as evidence that neither one can be trusted.
Thank you for listening. I’m Jeff Payne. I will see you next time.
I was 40 seconds from buying an egg peeler. Bundle selected, payment buttons on the screen, decision made. Then I did something I have started doing without really thinking about it — I asked an AI whether the storefront was legitimate.
It answered in seconds. And it never left the page to do it.
The Decision Was Already Made
This was not research. I was not comparing options, reading roundups, or asking anyone for a recommendation. I had moved past all of that. I was at checkout.
That is the detail worth holding onto, because it relocates the moment of scrutiny. For 20 years we have designed marketing around a buyer who investigates early and commits late. The verification step occurs in the opposite order: after the decision, at the point of purchase, and it is aimed at a specific page.
The check no longer happens at the top of the funnel. It happens with your hand on the button.
It Never Left the Page
What convicted that storefront was not an outside source. No scam database, no consumer report, no unhappy customer on a forum. The assistant read what the seller had published and compared it to itself.
The photograph at the top of the page (see screenshot below) showed a raw egg being cracked into a bowl.
The product underneath it was sold as a peeler for already-cooked eggs. Those are two different jobs.
The reviews contradicted each other on the basic mechanism — one praised a hand-cranked design, the next described dropping an egg into a container and shaking it.
And the page claimed a customer count in the six figures for a gadget nobody has ever heard of.
None of that required any knowledge of egg peelers. It required only the ability to read several things at once and notice that they cannot all be true.
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
Every Trust Signal Was Present. None of Them Mattered.
The page was not short on credibility markers.
Five stars.
A six-figure happy-customer number.
A secure-encryption badge.
A guaranteed-safe-checkout seal.
A 90-day money-back guarantee. A full row of recognizable payment logos.
Every signal a buyer was ever taught to look for was sitting right there, and every one of them passed a human skim. They were irrelevant, because coherence is a different test than credentialing. A badge is a claim about a process. Consistency is evidence about a story.
The Exposure You Built Yourself
Most reputation anxiety points outward: the review you cannot remove, the article that got it wrong, the third-party page that describes your business inaccurately. That exposure is real, and this show has covered it at length.
This is a different one, and it is entirely self-inflicted. It is the material you published yourself at different times for different reasons that no longer agree with each other. The About page written in 2019. The press release from the acquisition. The service list on the website that does not match the service list on the Google Business Profile. The bio carrying an old title. The case study whose numbers quietly contradict the case study beside it.
All of it was harmless as long as nobody read it together. A human buyer skims one page and never opens eleven of your properties side by side to compare the claims line by line. That is now a four-second job; it costs nothing and happens at the moment someone is deciding whether to call you.
Drift used to be invisible. Now it reads as a verdict.
The Verdict Is Not “Wrong.” It Is “Unreliable.”
Notice what a contradiction actually produces. It does not produce a finding that a particular fact is incorrect. It produces a judgment about the source: this company cannot be trusted to describe itself accurately.
That is a far stickier problem than an error, because it does not attach to a claim you can go correct. It attaches to you.
A buyer will forgive being wrong about something. Nobody hires a company they have decided is unreliable.
Credentialing Is the Easy Test
The awards page, the years-in-business line, the certifications, the badges — that is credentialing, and credentialing was always the easy test. It simply used to be the only one anybody ran.
The harder test has no owner inside most companies. Marketing owns the website. Someone else owns the Google Business Profile. The founder owns his own LinkedIn. The press release is four years old, and nobody has read it since the day it went out. Nothing in that picture is wrong. Nothing in it agrees, either.
The useful reframe is to stop treating a web presence as a collection of marketing assets and start treating it as testimony — because that is how it is being read now. Not one page at a time by a distracted human, but all at once by something looking specifically for the places where the story doesn’t line up.
The question worth sitting with is if somebody pointed a machine at everything your company has ever said about itself and asked whether it all holds together, what would it come back with? Not whether it would find something good — whether it would find agreement.
Proof over proximity still holds, but proof only counts when it corroborates. Two of your own claims that contradict each other do not cancel out to a neutral result. They read as evidence that neither one can be trusted.
The old standard was: is each piece good? The new standard is: do they all agree?
Frequently asked questions
What does it mean for an AI agent to cross-examine a business against itself?
Instead of checking a company’s claims against outside sources, an AI assistant can read many of that company’s own published assets at once — website pages, profile listings, reviews, bios, old announcements — and identify statements that cannot all be true simultaneously. The contradiction is found entirely within material the business published itself.
Why do trust badges and certifications matter less than they used to?
Badges and certifications are credentialing signals: they indicate that a process was completed. They say nothing about whether a company’s claims agree with one another. When a buyer runs a coherence check, a page can carry every conventional trust marker and still fail, because the two tests measure different things.
When does an AI verification check actually happen in the buying process?
Increasingly at the end rather than the beginning. Buyers who have already decided are pointing an assistant at a specific page immediately before purchase — closer to a final due-diligence step than to early research. That places the risk of inconsistency at the point of sale rather than at the top of the funnel.
What is the business consequence of inconsistent information across owned properties?
The finding is not that one specific fact is wrong. It is a judgment about reliability — that the business cannot be trusted to describe itself accurately. That conclusion attaches to the company rather than to a correctable claim, which makes it considerably harder to undo.
This episode is drawn from my own experience at the online checkout page for the egg peeler. The screenshot in my post shows that page and Gemini’s response to it, unedited except for the removal of the storefront’s name and domain, the two reviewer names, and the alternative product the assistant suggested. Those omissions are editorial — the point is the pattern, not the seller.
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
