For Example Dental Practice in Leicester, we asked 3 AI assistants 6 questions a customer might ask, 3 times each. The score blends how often you were named with how early in the answer you appeared. It is a snapshot, not a ranking.
Businesses named most often across the same questions. Names are extracted automatically from the answers, so check them before acting.
Sources cited across all answers (number of answers citing each). Your own site was cited in 14 answers.
Answers naming you, out of answers collected. No question mentioned your business name, so this measures whether assistants bring you up on their own.
Schema and llms.txt are tidy-up items, not proven ways to get named.
Your site was cited in some answers, a good sign. Clear service and town pages give assistants more to quote.
What to do: Give each main treatment its own page naming the town, the clinicians and price ranges, and answering patients' questions. Show name, address and phone on every page.
Business websites were cited in 90% of 159 UK AI Mode answers (Whito); businesses with a site were named 27% of the time vs 9% without (LocalPulse).
Google reviews or profiles feature in most UK Google AI answers. Bing is reported to feed ChatGPT search and Copilot, but no study has measured an effect from Bing Places.
What to do: Fill in every field (services, categories, hours, photos, booking link) and keep name, address and phone identical everywhere.
Google reviews or profiles were cited in 81% of 159 UK AI Mode answers (Whito). The Bing effect is plausible but unmeasured.
Businesses that AI assistants name tend to have far more reviews than those they skip.
What to do: Ask every happy patient for a review (a text link works well), reply to each, and never filter, incentivise or fake reviews.
UK study of 500 map-pack businesses: named businesses had a median of 147 reviews against 51. Correlation, not proof.
In this audit, answers leaned on demo-local-guide.test, example-directory.test, sample-street-dental.test, placeholder-smiles.test. Being listed accurately on pages AI reads is often easier than outranking them.
What to do: Claim and correct your listing on those sources and on the directories patients actually use for your type of business. Skip directories that do not apply to you.
Cited directories are sector-specific: Checkatrade was in 83% of trade answers but 1 of 79 professional-service answers (Whito).
Inexpensive hygiene items, listed for completeness, not because they are proven to change who gets named.
What to do: Add LocalBusiness schema, make sure robots.txt lets search crawlers in, optionally add llms.txt. Do not expect these alone to move results.
300,000 domains showed no clear llms.txt effect on AI citations; schema barely correlated with recommendation (Search Engine Journal; Insites).
We run these checks every month, do the work above and show you the movement. We cannot promise to get you named, only to do the things that correlate with it and to measure honestly.