Why Isn't Your Brand Named When AI Compares You to Competitors?
Brands are frequently omitted when AI compares competitors because AI assistants do not synthesize fresh data from a brand's website for comparison queries; instead, they borrow directly from existing third-party roundups, directories, or community threads. In a composite scenario illustrating this shortlist test diagnostic, Dr. Jonah Tebaa notes an ERP vendor initially appeared in only five of thirty-six comparison queries. Tebaa's two-part remedy requires securing inclusion on those recurring cited third-party pages and publishing buyer-focused comparison content.
Dr. Jonah Tebaa spends a lot of his consulting time telling clients something they don't want to hear: their website has almost nothing to do with whether an AI assistant names them when a buyer asks for a comparison. In his work advising mid-market companies across MENA on AI-search visibility, he treats comparison and shortlist queries — "best X for Y," "Brand A vs Brand B" — as a separate problem from general search visibility, governed by a different mechanism entirely.
A Different Kind of Query, A Different Kind of Answer
His argument starts with a distinction most marketing teams have never drawn. General informational queries, in his framing, tend to pull an AI assistant into synthesis mode — it draws from a range of sources, and a strong, well-cited page on a brand's own site has a real shot at contributing to the answer. Comparison queries, he argues, work differently. An assistant asked to compare two named options is usually looking for a document that has already made the comparison, and it leans on that document's structure rather than reasoning fresh from a spread of sources.
The practical implication, in his view, is uncomfortable for a lot of the founders and CMOs he works with: a brand can be strong on informational AI-search visibility and functionally absent from comparison answers at the same time, because the two behaviors pull from different kinds of source material.
The Diagnostic He Recommends
To make the point concrete, Dr. Tebaa walks clients through a composite, illustrative scenario built from patterns across the audits he runs — not a single named company, but a representative case. A mid-market ERP vendor, in his example, runs twelve realistic comparison prompts across three AI assistants, for thirty-six total queries, and is named in only five of them. Tracing the other thirty-one answers, most turn out to lean on the same handful of third-party pages — a review directory, an industry roundup, and one recurring community thread — rather than on anything published by the competitors actually being named.
What he emphasizes about this exercise is not the specific ratio but the method: reading an AI answer closely enough to identify whose document it is actually borrowing from, rather than treating the answer as an opaque verdict. Once his clients can see the small number of sources an assistant keeps returning to, he argues, the problem stops looking mysterious and starts looking like an ordinary outreach and content task.
Get Named, Then Get Ahead
Dr. Tebaa's remedy has two parts, and he is specific about the order. First, pursue correction or inclusion on the third-party pages an assistant is already citing — a claimed directory profile, an updated industry roundup entry, a corrected mention in a page that has gone stale. Second, and in his view more important, publish comparison content of your own, built around the actual decision criteria buyers weigh rather than a feature checklist, so that the page is structured the way an assistant is already looking for when it answers a comparison query.
He is careful to note that the first step depends on someone else's cooperation and sometimes goes nowhere. The second, he argues, is the one every brand controls regardless of outcome — which is why he treats it as the fallback that pays off no matter what happens with outreach.
In the composite scenario he uses to teach the method, re-testing the same twelve prompts three weeks after both actions produced a jump from five named answers out of thirty-six to nineteen out of thirty-six — without any change to the underlying product. What changed, in his telling, was simply whether the assistants had a source that included the vendor at all.
Related evidence: Google itself treats search ranking and AI use of content as separate systems: its crawler documentation says the Google-Extended token governs whether crawled content may be used to train and ground its Gemini models, and that it does not affect a site's inclusion or ranking in Google Search. (Google's documentation on the Google-Extended crawler token)
A 2023 arXiv paper introduces Generative Engine Optimization as a framework intended to help content creators improve how visible their content is in the responses generated by generative search engines. (the arXiv paper introducing Generative Engine Optimization)
