Answer: Yes, but much less often than big brands. One audit started from a complete list of every restaurant, cafe, and bar in two markets, 4,776 in all. Across 2,208 answers, 85.6% of those venues were never recommended by ChatGPT, Perplexity, Gemini, and Claude (Source: Pitenin, 2026). The ones that did surface were independent local venues, not chains.
That is the honest shape of it. Small businesses are named, the floor is low, and the reasons behind the floor are measurable rather than mysterious. What follows is what the published research actually shows, including the parts that are uncomfortable.
Do assistants actually name small businesses, or only chains?
They do name them. The best proof of it started with a full list, not a sample. It took every cafe, restaurant, and bar in two cities, 4,776 of them. It then checked 2,208 answers from ChatGPT, Perplexity, Gemini, and Claude, built from 96 questions a real buyer might ask. The plan was written down before the test began. The result: 85.6% of those places were never named by any of the four. Even among places with fifty or more ratings, 72.6% were never named (Source: Pitenin, 2026).
Turn that around and the other half of the finding appears. Roughly one venue in seven did get named, and the study population was independent local food and drink businesses, not national chains. So the flat version of the worry, that assistants only ever hand out famous names, is not what the data shows.
The caveat belongs right next to it. That study covers two neighbourhoods and one industry. Its value is the complete denominator, which almost no other study has, and a complete denominator is exactly what makes a number like 85.6% meaningful instead of a guess.
How much less often does a smaller brand get named?
Two larger studies put numbers on the gap, and they agree on the direction.
The first audited about 37,000 production runs across 215 commercially framed questions spanning 19 sectors, scored against a 533 brand reference list sorted into five levels of public prominence. Category leaders appeared in nearly every relevant answer. At the bottom two levels, the specialists and the regional players, 48 to 52% never surfaced in any of the 37,000 runs (Source: Jack, Lehman, Maloney, and Xu, 2026).
The second looked at more than 100,000 answers covering more than 100 brands tracked between March and May 2026, and found a three step ladder on a brand's first run: household names appeared in 73% of relevant answers, established mid market and regional brands in 44%, and niche or small brands in 11% (Source: Kumar, 2026). That study was published by a company selling visibility tracking, which is worth knowing when reading it.
Both numbers describe lists of brands the researchers chose, measured on the questions the researchers wrote. Neither is a prediction for any single business, and anyone quoting one of them back as your odds is misusing it.
Why do the big names win by default?
Because the AI leans on names it already knows. Someone tested that head on. Every product in the test had the same specs. The known brands took 100% of the picks (Source: Chu and Hou, 2026).
The same test found the interesting part. That total dominance disappeared once a rival held a rating advantage of less than a tenth of a star. The preference for the familiar name is real, and it is also shallow enough that a genuinely small difference in the visible evidence breaks it.
Hold that finding loosely. It ran on simulated product descriptions in a single category across three model families, so it is a demonstration of a mechanism, not a measurement of any real market.
What changes whether a small business gets named?
How the buyer asks, more than most people expect. One study used six AI models and five kinds of product. It compared plain questions against ones that spelled out what the buyer needed, 1,200 lists each way. Saying the goal out loud changed which brands came back. Brands missing from the plain answers could still be reached (Source: Malthouse, Lee, Yang, Pal, and Feng, 2026).
That is the opening, and it is a practical one. A business that is invisible for "best coffee shop" may be very findable for the thing it is actually best at, if the page says that thing plainly. Specifics are what a need shaped question can match: who it suits, what it costs, where it operates, what it deliberately does not do.
A separate controlled test points the same way on the page itself. Across 252,000 paired comparisons on six models, plain relevance to the question dominated everything else, with an explicit price and a recent date helping consistently (Source: Vishwakarma, Kumar, and Jamidar, 2026). That test compared two pages already placed in front of the model, so it describes which one an answer leans on, not how a page gets found in the first place.
Is this worth a small business bothering with yet?
The demand is real and it arrived fast. In BrightLocal's 2026 survey of 1,002 US adults, 45% said they had used AI to find a local business in the past year, up from 6% a year earlier in the same survey (Source: BrightLocal, 2026).
So both halves are true at once. Most small businesses are not being named, and a large and growing share of buyers are asking anyway. That combination is the whole argument for finding out where a specific business actually stands instead of assuming.
What can nobody promise about any of this?
Three things, and they rule out most of what gets sold in this space.
Nobody can convert a study rate into odds for one business. The tier figures above come from brand lists and question sets chosen by researchers, and a real business is not a row in one of those lists.
Nobody has shown a reliable way to climb the ladder. A survey of 45 studies on this exact question concluded that no reviewed technique shows a stable, long running, cross platform causal effect on being discovered (Source: Martinez, 2026). The honest version of the work is making a business easier to match and easier to verify, then measuring what changes.
Nobody can promise a number, a placement, or a date. Anyone offering one is describing something they cannot see.
Where does that leave a small business?
Knowing the floor is low is not the same as knowing where you stand, and the second one is measurable. We ask ChatGPT, Perplexity, Gemini, and Claude the questions your customers really ask, many times each, and count how often each one names your business and what it says about it. The first check is free.
Free. No card, no call.