Boris Dzhingarov, founder of ESBO Ltd, leads a global Digital PR & SEO agency dedicated to enhancing brand awareness and visibility.
getty
My agency now starts new client work with a check that did not exist in our process two years ago. We ask the leading AI assistants: Who would you recommend in this category, and what are the alternatives to the market leader? Then we record which brands get named and which sources the answers lean on.
The results rarely match the search rankings. Brands that spent years earning first-page positions in search are often missing from the AI’s answer entirely, while a competitor with a steady presence in comparison articles and community threads gets named again and again. The first few times, I assumed our test was flawed. It was not. A brand’s position on a search results page and presence in an AI answer are two different assets, and plenty of companies own only the first.
The Clicks Are Moving To The Answer Layer
The behavior behind this shift is well documented. The Pew Research Center analyzed the online browsing activity of 900 U.S. adults and found that when Google shows an AI summary, people click on a traditional result in only 8% of searches, roughly half the rate of search result pages without an AI summary. Adobe’s numbers point the same way from the other side: AI-referred traffic to U.S. retail sites grew 393% year-over-year in the first quarter of 2026, and that traffic now converts better than visitors from older channels. Fewer people are scanning 10 blue links. More of them are asking an AI model a question and acting on whatever it says.
For anyone whose living depends on brands being found, that adds a second scoreboard. A brand can rank first in search and still be absent from AI answers.
Where Assistants Get Their Opinions About You
The part many marketing teams miss is that AI models do not take your word for anything. When an assistant decides which brands to name, it leans on what is said on review platforms, in comparison pieces in the trade press, in forum threads where real users complain or praise and on reference sites. In the Pew study, Wikipedia, YouTube and Reddit were among the most cited sources in Google’s AI summaries. Our client checks show the same split: Independent pages that evaluate brands get cited constantly. The brands’ own pages get cited much less.
Consistency counts for more than most people expect. If your website, your LinkedIn page and a business directory all describe your business in different ways, the model has to guess. In my experience, it usually responds by hedging or by leaving the brand out entirely. Fixing contradictions is boring work. It is also one of the cheapest wins available in marketing right now.
What I Tell Brands To Do About It
Start by interviewing the AI assistants yourself. Once a month, pose the same questions to ChatGPT, Gemini, Claude and Perplexity: What does this company do? Who is it for? What are the alternatives? Save the answers, and note which sources they cite. The exercise costs nothing, and it shows you the distance between how you describe yourself and how you are described.
Then, try to get earned coverage within the sources that get cited. Digital public relations mainly used to be about links and rankings. The same placements now do a second job. A fair review in a trade publication or an expert quote in a respected newsletter is exactly what an assistant reads when it composes an answer. It also reads the comparison article that includes you. Pitching the journalist who writes the category comparison was worth doing two years ago. It is worth more today.
The monthly interview exercise gives you the target list, since the sources the assistants keep citing are the sources worth pursuing. Start with the ranking roundups in your category, including the “best of” and “alternatives to” articles your buyers already read. Find the writer or editor behind each one, ask what it would take to be considered for inclusion and make the decision easy: Provide accurate product details, access to the product itself, a data point they cannot get anywhere else or a customer willing to go on record. Do the same with trade publications that run reviews, and put your executives forward when journalists ask for expert comments.
Finally, add one number to your reporting: how often your brand appears when an AI model answers the questions your buyers ask. Rankings still matter, and I am not telling anyone to abandon search. But if the only thing you measure is position on a results page that fewer people read, you are measuring the past.
Two caveats belong here: Answers shift by model, by phrasing and even by day, so treat any single test as a snapshot rather than a verdict. And nobody can promise a brand a spot in an AI answer. Anyone who guarantees one is trying to sell you something. Earned coverage raises the odds the same way it raised the odds of landing on an analyst’s shortlist in every era before this one.
I will flag my bias: I have spent more than a decade getting companies covered, so of course I believe coverage matters. But the referral data said it before I did, and every check we run for clients keeps saying it: The brands that AI assistants recommend are the ones other people have written about.
Forbes Agency Council is an invitation-only community for executives in successful public relations, media strategy, creative and advertising agencies. Do I qualify?
