The figure comes from SOCi's 2026 Local Visibility Index, which looked at about 350,000 locations belonging to 2,751 multi-location brands and found ChatGPT recommended 1.2% of them. It is not a measure of small businesses, and SOCi, which sells local-marketing software, published it. It shows that AI recommends far fewer places than Google's local pack does. It does not tell you your own odds.
What SOCi measured
The index covers 2,751 brands and roughly 350,000 locations across 42 local industry categories, using more than 120 metrics across Google, Yelp, Facebook, ChatGPT, Gemini and Perplexity (SOCi, benchmark page published 12 Aug 2026, updated 9 Oct 2026).
| Platform, all industries | Share of locations surfaced or recommended |
|---|---|
| Google local 3-pack | 35.9% |
| Gemini | 11.0% |
| Perplexity | 7.4% |
| ChatGPT | 1.2% |
SOCi describes AI platforms as “3 to 30 times more selective than traditional local search”.
The sample is multi-location brands
Every location in the index belongs to a brand with several locations. A single-site plumber, dentist or café is not in it. The result describes how AI treats the branches of established brands. It may well be harder or easier for a one-site business; this data does not say.
The average hides a wide spread
SOCi's own breakdown of ChatGPT's recommendation rate: retail 10.2%, restaurants 5.3%, financial services 3.1%. All three sit above the 1.2% average. For the all-industry average to be 1.2%, other categories must sit far below it, or the average is weighted in a way the page does not explain.
Insight
1.2% describes the branches of multi-location brands across 42 categories. It is not a rate any particular business should expect.
What the 4.3-star average means
SOCi reports that locations ChatGPT recommended averaged 4.3 stars (Perplexity 4.2). That is the average of the recommended set. It does not show that 4.3 is a line below which a location is excluded, though many summaries treat it as one.
SOCi itself writes that reviews “work less like a ranking factor and more like a filter for inclusion”. That is its reading of observational data. It was not tested.
More people are asking, by vendor estimates
In SOCi's own 2026 Local Discovery Index survey, 52% of consumers had used an AI tool to research a local business in the last 30 days, up from 9% a year earlier. Other vendors publish different survey figures. Treat all of them as directional.
What the page does not tell you
SOCi's benchmark page does not explain how a “recommendation” was counted on any AI platform, how many prompts were run per location, or how the brands and locations were selected. It describes the index as independent but does not say how that was ensured. Until those are published, the rates cannot be compared with any other study.
What to do with it
For a small business, three things hold whatever the exact rate is. Ask the engines the questions your customers ask, not your brand name. Look at what they cite, and whether those sources describe you accurately. Ask again in a few weeks, because answers move. The method is in How to check what ChatGPT says about your business.