About KeenTag
KeenTag is an AI visibility tracker for ChatGPT, Perplexity and Gemini. When a buyer asks one of them what to use, the answer is a recommendation you never see. KeenTag asks those questions for you, reads every answer, reports how often you’re named as a rate with its 95% range, hands you one specific fix, and re-checks to show whether anything changed.
Four principles
They decide what we show, what we leave out, and what we refuse to claim. If a feature can’t keep all four, it doesn’t ship.
AI answers change from one run to the next, so a single number would be false precision. We report the share of answers that name you with its 95% Wilson range. Few answers give a wide range, and we show it wide.
We keep each answer in full and mark it with one of four labels. You can read every call we made instead of trusting a summary, and see which rivals were named when you weren’t.
Where rivals are named and you aren’t, we point to one page and one change a person can publish: a clearer first line, an answer-first passage. Never hidden text, never tricks written for machines.
/features/invoicingSay who it’s for in the first line.After you publish, we ask the same question again. We call it a change only when the new range no longer overlaps the old one. Otherwise we say there’s no detectable change yet.
How it works
The same loop runs on every plan. Larger plans track more brands and more questions. Every plan repeats the check on a schedule.
We read your homepage for your brand name and the category you compete in. That category is what buyers ask AI about.
Each question goes to every engine, more than once, because answers vary. Samples per question, per engine: Solo 3, Pro 3, Agency 4.
Named, not named, unclear or no answer. The rate is the share of answers that name you, and it always comes with its 95% range.
For a question you’re missing, the page to change and what to say, based on what the named rivals have that you don’t.
The same questions run again (Solo: weekly tracking; Pro: daily tracking; Agency: daily tracking). A move counts only when it clears the noise.
How the range is computed, what counts as named, how a re-check is judged, and what we keep and for how long.
Read the methodLines we don’t cross
A measuring tool is only useful if you can trust it when the news is bad. These are the promises that make that possible.
Your setup drafted from your site: up to 3 rivals, up to 5 buyer questions, 3 engines.