Answer engines don't rank pages, they extract answers. Paste a URL and see the questions your page answers, how well it answers them, and exactly which sections to fix.
Free tool; bring your own OpenRouter key. You pay only your provider (typically ~$0.01–0.15 per analysis).
ChatGPT, Perplexity, and AI Overviews don't rank your page, they break it into passages and cite the ones that answer a question directly. A page can be perfectly keyword-optimized and still never get cited, because its answers are buried, fragmented, or missing. Keyword tools can't see that. RelevancyLens audits your page the way answer engines actually read it.
Density, backlinks, SERP positions: signals from ranked-list search. Blind to whether a passage can be extracted as an answer.
Generate new words, but can't tell you what your existing page answers well, answers weakly, or fails to answer at all.
Maps your page to the questions it answers, grades every answer, and fixes the weak ones with AI, in place.
Every report is built from the same two questions answer engines ask: does this page answer things well, and does it stay on topic?
Answer Effectiveness (how well the page answers the questions it takes on) kept separate from Topical Cohesion, how focused it stays. Two different problems, never averaged into one vanity score.
The questions your page actually answers, each graded on the rubric answer engines care about: directness, completeness, self-containedness, and specificity.
Your page rendered as a flowing document with each section marked by role: which question it answers, or whether it answers nothing and dilutes the page.
The questions a reader on this page would ask next that your page never answers. Each gap comes with a one-click "Draft section" to close it.
Title, meta description, and H1 each scored against your page's real primary topic, with AI improvement a click away when they drift.
The concepts, tools, and names your topic implies, split into what you already cover and what's conspicuously absent.
Every finding in the report carries its own action. The page becomes your editor: apply AI suggestions in place, re-analyze to confirm the scores moved, then take the improved draft back to your CMS.
From URL to improved draft in one sitting
We fetch the page like an AI crawler, discover its primary topic, and score every section. Live progress, 30-60 seconds.
Answer Effectiveness × Topical Cohesion, the Question Map, section labels, gaps, entities, and metadata scores, with the worst problems surfaced first.
Apply AI fixes in place, re-analyze to confirm the scores improved, then copy or export the optimized draft into your CMS.
We'd score poorly if we didn't.
The tool is free and requires no signup. You bring your own OpenRouter API key, and you pay OpenRouter only for the model usage each analysis consumes, typically $0.01–0.15 per page depending on the model you pick and the page length. See the Model Guide for choices and costs.
Answer Effectiveness grades the answers your page gives against the rubric answer engines use to pick citations: directness, completeness, self-containedness, and specificity. Topical Cohesion measures, via embeddings, how tightly every section aligns with the page's primary topic. They diagnose different failures, a focused page can answer badly, and a page full of great answers can be unfocused — so we never blend them into one number.
Any model available on OpenRouter, you choose per task in Model Preferences. The default is an advanced open-weight model that keeps a full analysis in the cents range; you can switch to frontier models whenever you want a second opinion.
We analyze public URLs only and don't store your content, analysis happens in your browser session. Your API key lives in that session, never in our database. The one exception is deliberate: if you click "Save & share", that report is stored so your view-only link works. Recent analyses are remembered in your own browser's localStorage only.
Yes, save the analysis and you get an unguessable view-only link that shows the full report, including your applied improvements. Viewers don't need an account or an API key, and they can't spend yours.
One URL, 60 seconds, and a fix-list you can act on today.