Article·AI & Search
Why believability beats visibility now
Can people find us was the old question. Will an AI system trust us enough to recommend us is the new one, and vague positioning is what it filters out first.
By Jennifer Compton, CEO & Founder·
For years, the marketing question was simple: can people find us? Now there’s a second question just as important: will an AI system trust us enough to recommend us?
I read a study from Harvard Business Review that has changed how I think about brand building for our clients. Researchers asked ChatGPT, Claude, and Gemini for running shoe recommendations. Nike, the biggest athletic brand in the world, showed up inconsistently. Brooks, a much smaller brand, showed up reliably. The researchers’ explanation was that AI systems aren’t recommending brands based on heritage or awareness. They’re recommending brands whose claims are specific, structured, and backed by evidence they can verify. The researchers call this AI recall share, how often a brand gets pulled into the conversation when it actually fits what someone is looking for, and it’s one of the more useful ideas I’ve read all year.
Part of how Brooks earned that recall share is so interesting. Long before any AI recommendation existed, Brooks spent years teaching runners the vocabulary for their own problems, terms like overpronation and gait deviation, through coaches, specialty running stores, and running media. That vocabulary shaped the exact questions people would later ask an AI system, and Brooks was already positioned to answer them. They didn’t just build a better product, they built the language buyers would use to go looking for it.
This is a good follow on to the piece I wrote earlier this year on SEO, AEO, and GEO, the search, answer, and generative engine optimization layers of the buyer journey. That piece was about showing up, about being referenced instead of just ranked. This one is about what happens after you show up. Once an AI system finds you, what does it use to decide whether you’re worth recommending.
It’s really a lot like how a reporter vets a source. They ask for specific outcomes, they want to name a source instead of writing “a leading provider,” and they want data and third-party validation instead of self-description.
For our clients in professional, financial, and legal services, this matters more than most places, because your buyers already discount self-promotion. They’ve been trained by their own compliance functions to ask for proof. Add an AI system doing the same filtering automatically, at scale, before a human ever sees your name, and the cost of vague positioning goes up considerably.
The bar for what counts as credible hasn’t changed, but who’s checking it first has.
What to look at over the next quarter
- 01Audit your top five claims. Pull the language your firm uses to describe what makes you different. If the sentence could apply to your three closest competitors with the name swapped out, it’s not evidence.
- 02Build proof into the structure of your content, not just the narrative. A case study that says “we helped a client grow” is fine, but a case study that shows the starting point, the work, and the result with a number attached is evidence an AI system and a buyer can both use.
- 03Weight earned coverage over more owned content. A quote in a trade publication, an analyst citation, a third party summarizing your point of view in their own words, all of that reads as independent validation in a way your own website never will, to a person or a machine.
- 04Give your experts one consistent point of view across every channel they touch. AI models look for a signal that shows up more than once, in more than one place, said the same way. Consistency is what turns an opinion into a citable source.
Good PR people know this: you have to be the source, not just an option on a list. Ten years ago, the first reader of your case study was potentially a prospect sizing you up, or a reporter deciding whether you were worth a call. Today it’s just as likely to be an LLM deciding whether you’re worth a citation, before that person ever gets involved.
If you’re responsible for how your firm is perceived, here’s a thought to consider. If an AI system was asked right now to explain why someone should trust your firm over a competitor, what would it actually have to work with? Named case studies with real numbers attached. Media hits. A point of view an analyst or reporter has already quoted. Original data or research with your subject matter expert’s name on it. Bylines that answer a specific question rather than describe a general capability. You need all of this to close the gap.
We’ve been doing some early thinking with clients on exactly this: GEO audits of their content, pulling their ten most-used marketing claims, and scoring each one against three tests. Is it specific. Is it backed by something outside your own website. Would it survive with a competitor’s name swapped in. Most firms find two or three claims that pass and a lot that don’t.
The result is a clearer view of which claims deserve more investment and which are unlikely to earn trust in a world where AI is increasingly shaping buying decisions.