Why AI Finds You Only After Buyers Know Your Name

The test I ran on my own name

Last week I asked ChatGPT, Perplexity, Gemini, and Claude about Power Positioning and FAME. All four tied both to me. ChatGPT and Perplexity went further and called me a strong fit for an expert-led firm with a positioning problem.

Then I asked the question a real buyer would ask. “Who’s the best fractional CMO for an expert-led consulting firm with a positioning problem?” No method name, no hint.

ChatGPT named four other people. I wasn’t on the list.

That stung a little. It also matched something ChatGPT had told me in the first answer. It said it couldn’t find enough independent, third-party evidence about my work, and that testimonials on my own site don’t count the same way.

What a billion sources say about the gap

A week later, Chris Donnelly, co-founder of Searchable, shared the data behind his BrightonSEO talk, “Seen, but not cited.” His team analyzed 1.09 billion sources that AI answers drew from.

Three numbers stood out. Only about 3% of what AI reads comes from your own website. Around 31% comes from your competitors’ sites. And 66% comes from earned sources, things like editorial coverage, institutions, reviews, social posts, and forums.

The split that matters most is the one he kept coming back to. When someone asks about your brand by name, your site makes up around 46% of what AI reads. When they ask a category question without naming you, it drops below 5%.

That’s my test, at scale. Once buyers know you, you’re usually seen. Getting cited before they know you is the harder part.

Why your own site can’t close it

Your website does its best work after someone already knows to look for you. It answers “who is Michel Fortin” and “what is Power Positioning” well. It does very little for “who should I hire,” because on that question the engine is mostly reading what other people say.

I wrote about part of this in brand memory. A model’s sense of who you are comes from what it absorbed in training and what it finds when it searches. Both lean on sources you don’t own.

Which means a lot of the work that moves AI visibility looks a lot like old-school reputation building. PR. Speaking. Partnerships. Reviews. Being talked about by people who aren’t you.

I’ve seen this before, long before AI

None of this is new to me. It’s just measurable now.

When I started The Copywriters Board in 1998, it was my site. What built my name was what happened next. Gary Halbert, Clayton Makepeace, John Carlton, and Joe Sugarman joined and posted their own critiques. Their presence said more about the board, and about me, than anything I could write.

In 2004, I wrote the copy for John Reese’s Traffic Secrets launch. It brought in $1,080,496.37 in 24 hours. I didn’t tell that story first. John did, in a report he called The Million Dollar Day. That report traveled further than anything I published about it.

It’s still traveling. This month, a marketer named Matt O’Connor posted on X that Traffic Secrets was the first sales page to hold his attention all the way to the end, and credited me by name. In the replies, Sonny Lowry brought up my interviews with Gary Halbert, and Matt said he used to be in my copy forum, where copywriters were “tearing chunks out of each other.” Twenty-two years later, other people are still telling these stories for me, in public, where AI engines can read them.

Same with the stage. Clients would often introduce me as their copywriter at internet marketing seminars. A client vouching for you in front of a room does what no bio can. Today, the room includes the AI answering your buyer’s question.

Share of voice comes before share of mind

In the Triple-P Model, I read visibility as a chain. Share of voice, then share of mind, then share of action. Searchable’s numbers explain why the chain breaks where it does for so many experts.

Your site can win share of voice on named questions. But share of mind, being the name that comes up when a buyer describes a problem, gets built mostly off your site. If the earned layer is thin, you stay visible to people who already know you and invisible to everyone else.

That’s why I think PR, digital PR, and brand awareness work belong in the same plan as content and SEO now. They’re not a separate budget line. They’re how you get into the answer before anyone types your name.

How I’m working on the gap

Here’s how I approach it, starting with my own brand.

Step one: write down the unnamed questions. I list the questions a buyer asks before they know any names. “Who helps consulting firms fix their positioning?” “What does a fractional CMO actually do?” Then I check who the engines name for each one. That list becomes the scorecard. Chris makes the same point. Choosing the questions is where teams struggle most.

Step two: say who you help and what’s wrong, in plain words. On a named question, a branded method works. On an unnamed one, the engine has to match your page to the buyer’s problem. That’s Focus and Aim in Power Positioning. One buyer, one problem, stated the way the buyer would say it. The two-minute positioning test shows quickly whether you’re there.

Step three: earn proof from other people. In FORCEPS, the proof that carries the most weight off your site is Relational, Credential, and Social. Who vouches for you, what credentials others confirm, and what people say in public. It’s what I tell clients to build, and it’s what built my own name for decades without my planning it. Podcast invitations came every once in a while. I wrote guest pieces for other people’s newsletters. Lately I’ve put more into LinkedIn. But I’ve never worked on it on purpose, at a steady cadence. An engine telling me it couldn’t find independent proof is my cue to change that.

Step four: pick the earned sources your buyers’ answers actually use. Searchable found the mix varies by industry. Institutions made up about three quarters of earned citations in health. Publishers made up about half in retail. So I look at which sources show up in answers for my category before deciding where to pitch. There’s no point chasing coverage in places the engines don’t read for your market.

Step five: track the same questions over time. I rerun the same question set on a schedule, so I can see whether earned work moves the unnamed answers. An AI visibility audit is the longer version of the same check.

What I’d watch for

Vendor data deserves a caveat. Searchable sells AI visibility software, and I haven’t seen the full methodology behind the 1.09 billion figure. I’d treat the exact percentages as directional. The pattern matches what I saw on my own name, which is why I trust the direction.

And none of this replaces a clear position. PR without a sharp claim gets you mentioned for the wrong thing. The order still matters. Claim the position, then get other people to repeat it.

So here’s the question I’d ask about your own firm. If a buyer described your client’s problem to an AI tomorrow without knowing your name, who would it recommend?

How I wrote this: based on my own test results and experience. I used AI to help organize and tighten the draft, then edited and fact-checked it myself.

---

Frequently Asked Questions