What 4,291 AI answers say about which businesses get named
AI assistants name a specific business in 21% to 33% of answers, depending on the engine. We analysed 4,291 answers across ChatGPT, Claude, Gemini and Perplexity between 28 April and 10 August 2026, and separately tested 60 real businesses across 41 niches. Half of those businesses never appeared once.
Published 14 August 2026. Part of our AI visibility research. Our newer, larger study is the September 2026 study of 12,792 AI answers.
What we measured
How often does each AI engine name a specific business?
Perplexity and Gemini name a specific business in 33% of answers. ChatGPT does it in 21%, roughly a third less often than either. If you judge your visibility by what ChatGPT says about you, you are reading the least generous engine of the four.
| Engine | Answers | Named a business | Sources per answer | Named in first 200 chars |
|---|---|---|---|---|
| Perplexity | 1,070 | 33% | 9.9 | 80% |
| Gemini | 1,070 | 33% | 5.2 | 67% |
| Claude | 1,076 | 31% | 5.1 | 71% |
| ChatGPT | 1,075 | 21% | 1.8 | 88% |
How many sources does an AI answer actually cite?
Between 1.8 and 9.9, depending on the engine. Perplexity cited an average of 9.9 sources per answer. ChatGPT cited 1.8. That is a spread of more than five to one, and it changes the strategy: being one of several acceptable sources can be enough to get surfaced in Perplexity, while ChatGPT behaves much closer to winner-take-all. A business competing only for the ChatGPT answer is competing for a far narrower slot than one competing across all four.
How many businesses are invisible to AI search?
Half of them. We tested 60 real businesses across 41 distinct niches, from masonry restoration to orthodontics to staffing agencies, running the buyer questions their own customers ask before choosing a provider. 30 of the 60 never appeared once. 48 of the 60 scored zero on our visibility measure. The average business owned 9.4% of the answers in its own category, and the strongest performer in the set reached 40%.
The useful read on that is not that most businesses are losing. It is that most categories are still unclaimed. In a market where two thirds of answers name nobody and half of businesses appear nowhere, the cost of being the one that answers clearly is lower now than it will be later.
How this was measured
Answer sample. 4,291 answers collected between 28 April and 10 August 2026 across four generative engines: ChatGPT, Claude, Gemini and Perplexity. Between 1,070 and 1,076 answers per engine. Each answer is a response to a buyer question drawn from a real business category, not a synthetic prompt set.
Scope. Generative answers only. Google search results were collected over the same period but are excluded from this study, because a ranked list of links and a written answer are different products and comparing them side by side would be misleading.
Named a business. An answer counts as naming a business when the business name or a known variant appears in the response text. Sources per answer is the count of references the engine returned with its answer.
Market sample. 60 businesses across 41 distinct niches, tested between 3 June and 14 August 2026. Each business was run against the buyer questions asked in its own category. A business scored zero when it appeared in none of them.
Limitations. The businesses in the market sample are drawn from our own prospect and client pipeline, not a random sample of the economy, so they skew toward small and mid-sized service businesses in North America. Engine behaviour also changes over time. These figures describe a window, not a constant, which is why the date range is stated rather than implied.
This page is a dated snapshot and is not updated. The September 2026 study used different engines, questions and samples, so its figures are reported separately and never added to these.
Find out where you show up
The free scan runs your business against AI engines and shows you which buyer questions you appear in and which ones a competitor owns.