AI visibility self-test: twelve prompts for your own brand
Fill in three fields and get the prompt set plus a scoring sheet — the measurement discipline stays yours.
Nothing leaves your browser. What you type here goes nowhere: no server call, no analytics, no storage. The prompts and the downloadable sheets are built on your own machine, and become files only when you click download.
The twelve prompts split into two blocks, and the order matters. The first six ask WITHOUT your brand name: that measures the commercially interesting thing — whether an assistant thinks of you when nobody has prompted for you. The second six add help, first by category and then by name: that separates "does not know you" from "knows you but files you elsewhere". The scoring sheet comes with it because measurement discipline matters more than the prompt list: ask each prompt at least three times, in separate conversations, and record a proportion rather than a yes or no.
Fill in all three fields and the twelve prompts appear here.
How to use it
Fill in the fields, copy the prompts, and ask them one at a time — in separate conversations, because an earlier prompt influences the next. For each one, record how many runs out of how many you appeared in and how: named, described without naming, or cited as a source. The downloadable sheet is already laid out for that, with a default of three runs. When you are done, keep the file unchanged: the next round has to use the same set, or there will be nothing to compare against.
Frequently asked questions
- Do you store what I type?
- No. There is no server call and no analytics on this page: the prompts are assembled in your browser and become files only when you download them. You can verify this by reading the page's source.
- Why twelve?
- Six unaided and six aided — enough for the difference between the blocks to be readable, and still runnable by hand in one sitting. The number is not the point; having both blocks is.
- Why ask each prompt three times?
- Because these systems are not deterministic: the same prompt can return different answers run to run. One run is a sample, not a result. Three is the minimum at this free tier; it separates the obvious cases, and finer changes need more.
- Does it work without JavaScript?
- The explanatory content and the methodology do — the page is server-rendered and readable. The generator itself needs JavaScript because it deliberately runs on your machine: that is the price of no data reaching a server.