Prompt Discovery reads your page and lists the realistic prompts it already answers well, the ones it is a paragraph away from winning, and the ones it misses. Every prompt comes with the specific change that would move it up. It is a read on your content, not a ranking check — we do not query ChatGPT or Perplexity to get these.
From one product page. The gaps are not exotic — they are what a buyer asks before deciding.
The questions come from what the page is about, so they fit the page you actually have.
The content as a crawler receives it, so the analysis is grounded in what is really published.
What someone arriving at this page wants to know — including the practical ones teams stop noticing.
Answered, partly answered, or a gap, with the reason. The gaps become the brief.
runs per month on Plus — 90 on Pro, 300 on Agency
run produces the whole question set for a page, not one question
keyword research required — the questions come from the page
Run it on any page. You get the question set, graded, with the gaps ranked so you know what to write first.
No, and nothing that claims to be is telling you the truth — prompts typed into an assistant are private and no volume data for them exists. These are the questions your page is positioned to answer, derived from what it actually says.
By reading it. The analysis works from what the text says and how it is organised, not from keywords, metadata or term frequency. A page that mentions a topic ten times but resolves nothing about it is positioned worse than one that answers a single question properly.
What the page is doing on that prompt today. Already strong — it answers this well. Close to winning — most of the answer is there. Needs content work — the page touches the topic without resolving it. Gap — a question people really ask that the page does not answer.
Close to winning. Those are pages that already did the hard part and are one addition away, so the work is small and the movement is visible. Gaps are the plan for next month, not this afternoon.
No. Each recommendation names the specific thing missing on that prompt and where it belongs. Where an FAQ block genuinely is the answer, it says so — but bolting questions onto the bottom of a page is what people do instead of fixing the page, and it rarely works.
Yes. Optimizations drafts it from your existing content, so a recommendation becomes something you paste in and re-audit rather than a writing task.
No. There is no volume here and no ranking. A tier describes the state of your page on that prompt, not demand for it or how any particular model would rank you.
One analysis of one page. Re-running an unchanged page returns the saved result rather than spending another.
Then the problem is upstream. Positioning only pays once a crawler can reach the page, read it from the raw HTML and get a response in time. Check those first — a well-positioned page nobody can fetch stays invisible.
Discovery starts from your page and finds the questions. Fit starts from a question and tests the page against it. Run Discovery to build the list, then Fit on the ones that matter.