Home › Services › AI visibility

Service

When your buyers ask AI who to call, does it name you?

Buyers now ask ChatGPT, Gemini, Perplexity and Google's AI for a recommendation instead of scrolling a results page. We ask those assistants the questions your buyers ask, without using your name, and show you whether you came up, who did instead, and where the assistants got their answer from.

Run the free check The full report, £9

Live calls to five assistants, not estimates. Results on screen and in your inbox in about five minutes.

At a glance

  • The free check puts five of your buyers' questions to ChatGPT, Claude, Gemini, Perplexity and Google AI. Twenty-five answers, each scored: you were named, a competitor was named instead, or nobody was.
  • The full report runs 34 questions across three buying stages, plus six branded controls. 170 answers, every competitor named, every source cited, the raw text of every answer. £9.
  • Nothing is simulated or extrapolated. You see the questions, the answers and the sources, so you can check the working.
  • Assistants name the businesses they keep meeting in the sources they read. The fix is structure, structured data and question-led content, on your site and on the sites they cite.
  • Nobody can guarantee a citation. What we can do is measure it, change the inputs, and measure it again.

Start with the free check

Give us your business name, your website and a sentence on what you do. We read your site, write five questions the way a buyer would type them (two describing the problem, two asking what to use, one asking for the best providers), and put each one to five assistants with web search on where the assistant offers it. None of the questions uses your name.

A classifier reads each answer and records whether you were named or your site was cited, which other providers were put forward, and which sources the assistant used. You get a 5-by-5 grid, a count per assistant, and a short league table of who was named instead of you.

Run my free check

The full report

The free check is a snapshot. The report is the measurement. It runs 34 unbranded questions, 14 describing the problem, 14 asking for a product or solution and 6 asking for comparisons, plus 6 branded controls to confirm the assistants can find you at all when asked directly. 170 scored answers across five assistants. It shows:

  1. Every competitor the assistants recommend instead, how often, and your share of voice against them.
  2. Every domain the assistants cited, most-cited first. This is the map of where you need to be published.
  3. The buying-stage gap. Whether you appear when buyers describe the problem, when they ask for a solution, or only when they already know your name.
  4. How often each assistant answered from memory rather than the live web, because your name only gets into those answers once it appears in the sources the models learn from.
  5. All 34 questions and every raw answer. The questions are yours to keep. They are the content brief.

Free check

Five questions, five assistants, about five minutes.

£0
No card, no call
  • 25 live answers, scored
  • Named / competitor / nobody grid
  • Who was named instead of you
  • Result on screen and by email
Run the free check
The measurement

Full report

34 questions, 170 answers, the sources and the gaps.

£9
One-off. Repeat it in four to six weeks to measure change.
  • Everything in the free check
  • Competitor league and share of voice
  • Every cited domain, ranked
  • Buying-stage and memory-versus-search gaps
  • All 34 questions and raw answers, printable
Get the full report for £9

Payment is taken by Stripe. The report is generated from live calls after payment and usually arrives within fifteen minutes.

What moves the numbers

The assistants can only repeat what has been published. Three things, in order.

  1. Answer the questions on your own site. Each question in the report with no answer of yours in circulation is a page you don't have: question as the heading, direct answer in the first paragraph, source, caveat, author, date. Start with the questions a competitor won.
  2. Make the business one consistent entity. Same name, same one-line description, same links on your site, LinkedIn, Google Business Profile, Companies House and every directory. Behind that, crawler permissions, a sitemap and structured data describing the business, its people and its services, so a machine gets an unambiguous account of who you are.
  3. Carry the same answers onto the sites the assistants cite. The domain table in the report is where they already look. The same claim, in the same wording, appearing on several of them under your name is what turns a mention into a recommendation. Native answers on each surface, linking back to yours, not syndicated copies of one press release.

We can do this work for you, or set it up so your team can. It is priced after the report, on a call, because the report tells us how much of it you need. You keep the report and all 34 questions either way.

What we will not claim

Nobody can guarantee a ranking or a citation, and anyone who does is selling something they cannot control. Assistant answers vary run to run, and the assistants re-crawl on their own schedule. A report is a snapshot of one day, not a verdict, which is why it uses 34 questions rather than five and why we recommend repeating it after four to six weeks to measure change. What can be said with confidence is that a business the assistants never meet has no chance of being named, and that fixing that is mechanical, durable and improves the site for human readers too.

How it fits with the rest

Visibility brings the enquiry. Chat and voice make sure somebody answers it, and quoting turns it into a price the same day. Doing the first without the others simply moves the leak further down the line.

Straight answers

Questions about AI visibility.

Is this just SEO with a new name?

No. Search ranks pages; assistants write an answer and name a handful of businesses. The assistants name the businesses they keep meeting in the sources they read. The check shows whether you are one of them yet. Titles, internal links and technical health still matter, but what is added is structure and content built around whole questions rather than keywords.

Why does the check take a few minutes?

Because they are real calls. Twenty-five answers from five assistants with web search on, then a scoring pass. We do not cache or estimate. The full report is 200 calls and usually lands within fifteen minutes of payment.

Will the result be the same tomorrow?

Not exactly. Assistant answers vary run to run, which is why the full report uses 34 questions and why we recommend repeating it after four to six weeks to measure change rather than reading one day as a verdict.

What is a "mention"?

Your name appearing in the answer body, in any spelling, or your domain appearing in the sources the assistant cited. Competitors are the other providers the answer puts forward as options, extracted by a classifier and spot-checked by a person.

What happens to my details?

They are used to run the check and email the result. The questions and answers are kept so you can revisit them. Nothing is shared or sold.

Do we need to write a blog?

Something like one, yes. The answer layer is the part that compounds, and roughly seven or eight articles in ten should answer real customer questions. The report hands you the first 34. We can write them from your material, or set it up so your team can.

Does this work for a local services business?

Often better than for a national one, because there is less competition for the specific questions people ask about your trade and area. Location pages only earn their place where you have genuine work and evidence in that area.

Find out what AI says about your business today.

Five minutes, no card. Then decide whether you want the full picture for £9.

Run the free check