✦ We optimize brand visibility across AI search and chatbots ✦
We scanned 87 small-business websites across ChatGPT, Gemini, Perplexity and Claude using the buyer prompts their own customers would type. Nearly half of them do not exist to AI.
43 of 87 businesses received zero mentions on any of the four assistants.
60 of 87 businesses were never reliably named for their own buyer prompts.
In 97 of 100 sampled ChatGPT buyer prompts, the business itself was not named.
344 competitor recommendations were surfaced across those same 100 prompts.
Tiers are assigned from total mentions across all four assistants.
Zero mentions across all four assistants. The business does not exist to AI.
One to three mentions, usually on a single assistant, never consistently.
Named occasionally, but competitors are recommended first in most answers.
Named in a meaningful share of buyer prompts on at least two assistants.
Reliably named across multiple assistants for core buyer prompts.
Named first or near-first across assistants, with competitors trailing.
Share of businesses that received zero mentions on each assistant.
| Assistant | Businesses with zero mentions | Zero-mention rate | Total mentions in sample |
|---|---|---|---|
| ChatGPT | 58 of 87 | 67% | 245 |
| Gemini | 68 of 87 | 78% | 118 |
| Perplexity | 72 of 87 | 83% | 100 |
| Claude | 86 of 87 | 99% | 25 |
of all mentions were captured by the top 10% of businesses.
median mentions per business across all four assistants.
mentions earned by the single most visible business in the sample.
Nearly half of the businesses scanned had zero mentions on any assistant for prompts describing their own service, in their own service area. These are operating companies with websites, reviews and years of trading history. The models simply had nothing to say about them.
99% of businesses had zero Claude mentions, and Claude produced only 25 mentions in the entire dataset. Assistants that browse less and rely more on training-set entity knowledge are the hardest to enter — and the slowest to correct once you are missing.
The top 10% of businesses captured 61% of all mentions in the sample. The median business had one mention across four assistants. AI answers do not distribute attention the way a page of ten blue links does — they name two to four options and stop.
Across 100 sampled buyer prompts, assistants named 344 competitors by name — an average of 3.4 per answer. Absence is not neutral. Every unanswered buyer prompt is a warm referral delivered to someone else.
Several businesses in the sample rank well in traditional search and still score invisible. Assistants resolve entities, not pages: they need consistent structured facts, corroborating third-party sources, and machine-readable positioning before they will recommend a name.
Published 2026-07-30 by Monic AI Systems. Cite as: Monic AI Systems, The 2026 Small Business AI Visibility Report. Journalists and analysts are welcome to reference these figures with a link to this page.
Run the same scan we used for this report against your own buyer prompts.