Gemini names companies less often than ChatGPT or Claude do.
Across the same set of prompts, the three engines named a cohort company at noticeably different rates, with Gemini the least likely of the three to name anyone from the cohort.
33% vs 42.8%
Each of the three engines in this study, ChatGPT, Claude, and Gemini, was asked the same 22,875 prompts and checked against the same cohort of 153,251 companies. The rate at which each engine named at least one cohort company in its answer varied measurably: ChatGPT named a cohort company in 42.8% of its answers, Claude in 40.5%, and Gemini in 33%.
That gap, roughly ten percentage points between the highest and lowest, means a buyer's choice of assistant is itself a variable in whether any given company gets a fair hearing, independent of anything the company has done. A company that has done nothing differently this month could look meaningfully less visible simply because more of the traffic checking it happened to land on Gemini rather than ChatGPT.
What this number can and cannot tell you. It cannot tell you why Gemini names a cohort company less often. There are at least two different mechanisms that would produce the same observed gap. The first is a stricter or more conservative answer style, where Gemini's underlying model or product configuration is tuned to name fewer specific businesses and rely more on generic category description. The second is a narrower retrieval surface, where Gemini simply has less specific company-level information available to draw on for a given prompt, independent of any deliberate conservatism. These would look identical in this study's appearance-rate figure but would call for different responses: the first is closer to a permanent style difference to plan around, the second is closer to a temporary information gap that might close as retrieval improves.
Distinguishing the two would require comparing appearance rates on categories where the underlying set of qualifying companies and their web presence is already well documented and roughly equal in quality, and checking whether the rate gap persists there too. This study did not run that isolated comparison this month, so the honest position is that the mechanism behind the gap is not yet known, only its size.
What follows practically. A company that depends heavily on being found through AI-assisted research should not assume all three engines are interchangeable channels. If its own buyers skew toward one assistant, that is the engine whose behavior matters, and general-purpose AI visibility claims that average across all three can obscure a real problem, or a real strength, on the one engine that counts for that company's specific market.
Other findings
What share of companies actually get mentioned by AI assistants at all?
Almost every company shows up somewhere. That is not the same as showing up where it counts.
Nearly all of the cohort was named by at least one engine during the study window, but that headline figure hides how unevenly that presence is distributed.
If ChatGPT recommends a company, does Claude or Gemini recommend the same one?
The three engines agree with each other on almost nothing.
Out of every pair of engine and company where at least one engine made a mention, the three engines landed on the same answer together only a small fraction of the time.
Once a company is mentioned by an AI assistant, where does it typically land in the list?
Being named is not the same as being named first.
The average position of a mention across all three engines sits in the middle of a typical answer, and the average shifts slightly depending on which engine is asked.
Want to know how AI answers describe you?
We run the same measurement on your category. Fifteen minutes with founder Omar Jenblat, your own numbers, no deck.
- Your category measured the same way
- Your own numbers, not a sample deck
- Fifteen minutes, no obligation