The self-check
Think of one page, such as your main service page. Answer honestly; "Not sure" counts as no.
How the score works
- Simple points. Each answer scores 2 for yes, 1 for partly and 0 for no or not sure, so the total is out of 16.
- Three bands. Thirteen or more is green, eight to twelve is amber and seven or less is red.
- Nothing hidden. That is the whole sum; there is no hidden weighting and no benchmark against other businesses, because we do not have one we could stand behind.
The eight questions are the basics that make a page easy for an AI assistant to read, trust and quote. They are the same areas a careful editor would look at:
- Is the answer near the top?
- Is each section about one thing?
- Are the facts written down and consistent?
- Is it clear who stands behind the page?
- Can machines read it?
Why these eight questions
Assistants that look things up build an answer from passages they can lift and trust.
- Something to lift. A page that opens with a plain answer, keeps one question per section and states its facts in sentences gives them something to lift.
- Easier to trust. Matching facts across your site and other profiles make the passage easier to trust.
- Easier to read. Structured data and an open robots.txt make it easier to read in the first place.
Our guide on how to write a page AI assistants can quote goes through each one with a before-and-after example.
The last question, about other sites, is the one you control least. It is in the list because assistants often rely on what directories, trade bodies and the press say, not only on your own site.
If a rival is named and you are not, our page on why AI recommends your competitor shows how to compare the two, question by question.
What to do with the result
- Work down the fix list in order.
- Write down the questions your customers ask (the question set builder drafts ten in a minute).
- Test real answers: ask them several times and record who is named.
- Re-check the same way each month.
A higher self-check score does not guarantee anything, but a page that fails the basics is unlikely to be the one quoted.
Honest limits
- A self-check, not a scan. This is built on your own judgement, not a scan of your page.
- Page analysis reads the page. GEO Agency's page analysis does read the page and scores it across seven areas with reasons and suggested fixes.
- Estimates only. It runs on Google Gemini, so its scores and answer simulations are estimates, not a record of what any assistant told a customer.
- No promises. Nobody can promise a place in an AI answer, and we do not.
Common questions
Is this score a measurement of my page?
No. It is a rough self-check based on your own answers, and the scoring is the simple sum shown on this page. It tells you which basics might be missing, not whether assistants will name you.
Is anything I choose sent to you?
No. Your answers stay in your browser and disappear when you leave the page. If you have accepted analytics cookies, we count that the check was started and the score button pressed, nothing more.
Why does "Not sure" count as no?
If you are not sure whether a page has a clear answer or matching facts, an assistant reading it may not be sure either. It is worth checking.
What would a proper check involve?
Asking assistants your customers' questions several times and recording who is named, and scoring the page itself. GEO Agency's page analysis scores a page across seven areas with reasons and fixes; it runs on Google Gemini, so the results are estimates.