How it works
How we check your AI visibility.
Read the answers behind the result and understand what they can tell you.
We ask questions a customer might ask
We ask AI assistants about your services and city. These include questions such as “best dentist in Oshawa”, requests for a nearby service and questions about your contact details. We keep the original answers and any source links returned. Your report names the assistants checked and the assessment date.
We count how often your business appears
If your business appears in 3 of 12 completed discovery answers, your mention rate is 25%. Discovery questions ask the assistant to suggest businesses without naming yours first. Questions that already include your business name are kept for checking facts, but do not count toward this percentage.
A failed or unavailable answer is shown separately. It does not count as an answer that left you out. The report shows how many answers were collected so you can judge how complete the result is.
AI answers can vary
The same question can produce different answers. We can ask each question one to five times per assistant and show a range around your result. A wide range means more uncertainty. Repeated answers may be related to each other, so the range is an estimate, not a prediction of what every customer will see.
For a useful comparison, keep the questions, assistants and settings the same. Changes to an assistant can affect results even when your website has not changed. A change between reports does not prove that your marketing work caused it.
You can inspect competitors and sources
We identify business names in discovery answers and count how often they appear. We also show links supplied as sources. Names are detected automatically and may need checking against the original answer. A source link gives your team something to investigate; it does not prove why an assistant chose a business or whether you have a listing on that site.
Recommendations follow the recorded answers
Findings can include an incorrect phone number, a missed service question or a website instruction that blocks an AI search tool. We explain what was observed and what your team should check next. A report is not a full audit of your website, business listings or booking system.
Website-access checks read robots.txt, the file that tells automated tools which pages they may access. Permission to read a page does not prove that it has been indexed or will be recommended.
Results may differ from the apps you use
We use each assistant’s business-facing connection, called an API. Its answers may differ from the public app, which can use a person’s location, history or different settings. Your report records this assessment, not every possible version of an assistant’s answer.
Technical details: statistical ranges and older scores
We use a 95% Wilson interval for discovery mention rates. Each successful answer is one observation. Because repeated AI answers may be correlated, this is an approximate description of sampling uncertainty.
Answer share is your detected business appearances divided by all detected business appearances. It differs from mention rate because one answer can name several businesses.
Older reports and the sample report can show a combined score across the areas below. Only areas with enough completed checks contribute; their weights are adjusted to total 100%. This score is different from the percentage of answers that mention you. Do not compare combined scores built from different sets of checks.
- Your details match everywhere (20%) — Your name, phone and address are the same on your website, Google and directories.
- AI can read your website (20%) — Your site loads, isn't blocked, and its main content is easy for AI to read.
- You answer what locals ask (20%) — Your site mentions your city and answers the questions customers actually ask.
- Your business info is labelled (15%) — Structured data describes your hours, services and location in a format software can read.
- Other sites back you up (15%) — Reviews, listings and other websites confirm you're a real, active business.
- AI mentions your business (10%) — When we asked AI tools real customer questions, did they name you?
A score marked “provisional” is an early estimate based on incomplete checks. Our score bands are internal labels, not an industry standard:
- Needs Improvement: 0–39
- Developing Visibility: 40–64
- Strong Visibility: 65–84
- Excellent Visibility: 85–100
We record the scoring-method version so a change in calculation can be distinguished from a change in the collected answers.
What the report does not promise
AI visibility is a snapshot. A report cannot guarantee future recommendations, search rankings, enquiries or sales. The example report uses a fictional business to show the format. See pricing and what is included before requesting your own report.