Pages
Audit your own pages for citability — one score, four evidence checks, and whether engines actually use the page in answers.
Pages is the audit surface for your own site. Add the URLs that matter, and each is fetched and scored on whether an AI engine would quote it.
Everything else measures the outside world. This screen measures the part you control.
The scoring model
One score, built from four checks:
| Question | |
|---|---|
| Citability | How quotable is this page for an AI engine? A single 0–100 score: the share of the four checks the page passes. |
| Evidence checks | The four ingredients the score is built from, each pass/fail with a note: concrete statistics, quotable expert or customer quotes, claims that cite sources, and a direct answer near the top. |
These four are the only content tactics with experimental support for AI visibility, which is why the audit measures exactly them and nothing softer.
A note on vocabulary, because it comes up: readable, citable and crawlable are not three separate scores here. A page an engine can read cleanly is a page it can cite — the two travel together, so they collapse into the one citability score. Whether an engine can crawl the page at all (robots rules, response codes) is a different, technical question this audit assumes is already solved; check it separately with the free AI Crawlability Checker.
Rendering is the one crawl question the audit does answer, because it decides what there is to score. The audit reads the HTML your server sends, before any JavaScript runs, which is exactly what ChatGPT's and Perplexity's crawlers read: they run no JavaScript. Cookie banners and other overlays your scripts add are not part of that HTML, so they are not part of the audit. A page whose text only appears once its scripts run is not scored; its audit fails with Content needs JavaScript and both character counts, because to those crawlers it is an almost empty page, and rendering it on the server is the fix that comes before any of the four checks.
The page table
| Column | Reading |
|---|---|
| Page | Title and URL. |
| In answers | Used — times the page appeared among an answer's sources; cited — times an engine quoted it against a claim. Measured from your collected answers, not estimated. |
| Citability | The audit score. |
| Evidence checks | Passed checks out of four. Hover the count for the pass/fail detail and the auditor's note on each. |
| Last audited | When the page was last fetched and scored. |
The two halves answer different questions on purpose: the audit says whether a page deserves citations, the answer counts say whether it gets them. A high-citability page with zero appearances is a distribution problem; a low-citability page already in answers is the fix that pays fastest.
Managing tracked pages
- Add page — paste a URL; it is fetched and audited against the evidence checklist. Start with the pages you want cited: pricing, comparisons, category guides, docs.
- Re-analyze — re-run the audit after shipping a fix, rather than waiting for the next pass.
- Delete — stop tracking a page. Its audit goes with it.
Track pages with a job to do. Fifteen well-chosen pages produce a better work queue than a hundred that nobody will ever edit.
Reading the scores
Compare across your own pages before comparing against any absolute standard. The useful questions are relative:
- Which of my pages fails checks and already appears in answers? Fix that first — it is being read today.
- Do my comparison pages score below my blog posts? Comparison pages are disproportionately cited for high-intent prompts; a gap there costs more than the number suggests.
- Did the score move after the last deploy? Re-analyze after each change so you learn which fixes actually pay.
From score to fix
Every failed check is an issue with a known fix — add the statistic, quote the customer, cite the source, answer the question in the first screen. The loop is: audit a page here → ship the fix → re-analyze → watch the page's used and cited counts, and its citations in Sources.