What do AI assistants tell families about your service?
Families, commissioners and job candidates now ask ChatGPT, Gemini, Claude and Perplexity for recommendations before they ring anyone. An answer gets produced, a few providers get named, and the ones that are not named never find out it happened. It does not appear in your analytics.
Four assistants, asked the questions your families ask
We do not check whether you appear when somebody searches your own name. We ask the questions a real enquirer asks — the kind that name no provider at all — from the right location, and record who gets named.
ChatGPT
The most widely used assistant.
Gemini
Reaches users through Google surfaces by default.
Claude
Used in professional settings.
Perplexity
Answers with citations, so sources matter more.
01Question set built to your service
Your registered activities, the conditions and age ranges you take, and your actual catchment — turned into the non-branded questions an enquirer would type.
02Live testing across the four assistants
Each question put to each assistant and the response recorded, so you can see the wording, not just a score.
03Comparison with nearby providers
Who is being named instead of you in your area, and what those providers have in the record that you do not.
04Hand verification
Every finding is checked by a person before it reaches the report. Assistants produce confident wrong answers, and an unverified figure is worse than no figure.
05Fix plan, then the fixes
What to correct and in what order — then we carry out the work rather than handing you a list.
Something you can hand to a board, not a dashboard login
- The verbatim answers. What each assistant actually said about your service, and about the providers named instead.
- A named-or-not result per question, so the finding is a fact rather than an impression.
- The nearby-provider comparison, showing which competitors are being recommended and on what basis.
- Where the record disagrees with itself — your CQC registration and rating, Companies House entry, local authority listings and Google Business Profile. An assistant resolving a conflict trusts the regulator over your website.
- A prioritised fix plan, separating what is wrong and must be corrected from what is merely absent.
- A dated scan reference, so the same questions can be re-run later and the two compared.
Ex VAT, quoted per organisation rather than per site. The check is a fixed fee; the remediation work is quoted separately once you have seen what needs doing, and you are free to do it yourself.
Nothing written here can outrun what you are registered to do
Care marketing carries a limit other sectors do not. You cannot claim outcomes you cannot evidence, and anything written to win a citation still has to be defensible if a family quotes it back during an inspection. So the work is written against your registration and your real capacity. Specific, checkable statements happen to be exactly what assistants cite, which makes being accurate and being found the same job.
If what the assistants repeat turns out to be a governance problem rather than a marketing one — a tool writing your content, or a record you cannot evidence — that is a different piece of work, and it is covered by Clinical AI Governance.
No client work is published or used in our marketing without written consent, and clients choose whether to be named or anonymised.
Find out what is being said before you spend anything.
Thirty minutes, no charge. We will tell you whether this is worth doing for your service, including when the answer is that it is not.
