Site Health

A Site Health crawl captures pages from your website and analyzes their technical and content properties. Begin with Run new crawl, then review the crawl’s scope, progress and coverage. Recrawl reruns the shown crawl with the same scope and settings, so Changes compares like with like.

Read the score with its coverage

Web summarizes scored website checks. AEO summarizes scored checks across answerability, structure, evidence, machine readability, provenance, freshness and crawlability.

A higher score does not promise rankings, indexing or citations.

A Partial audit can still show a score. Read the accompanying coverage to understand how many pages were scored and which applicable checks remain unresolved.

Open the affected page

For a finding you intend to fix:

1. Open its affected pages.
2. Read the captured evidence and page classification.
3. Check whether the issue applies to that page’s purpose.
4. Compare with the current website before making changes if the crawl is old.

Page classification affects which checks apply. A product page and an editorial article do not have identical expectations. An unresolved page purpose remains unresolved.

Scored and advisory checks

Only scored checks move the Web and AEO scores. They are the checks that affect whether search and AI engines can reach, read and trust a page, such as HTTPS, indexability, crawler access, valid structured data and visible attribution.

Other checks are unscored: advisory checks such as meta descriptions, Open Graph tags or content recency, and defects such as broken internal links. A failing unscored check still raises an issue you can work on; fixing it does not change a score. A few diagnostic readings, such as response time, are reported without raising an issue.

Understand how results roll up

Page scores use the scored checks that apply to the page. Checks marked unknown, error or partial do not receive passing credit. They remain visible in completion and coverage.

Crawl scores average the finalized page scores of the pages CiteLadder monitors for the project. A page with many checks does not automatically outweigh one with fewer checks. Avoid rebuilding the score from a filtered issue table.

Review crawler permissions

The crawler panel groups bots by purpose: search engines, AI search, AI training and user-triggered fetches. Filter by purpose to focus the table. Root access reports the homepage permission. Policy reports All allowed, Restricted, All disallowed or Unknown access over the crawl’s bounded URL sample. These can differ when rules allow the homepage but block other paths.

The matched group shows a bot-specific group, wildcard fallback or Not specified. Missing or unreadable evidence stays unknown. These observations describe robots.txt permissions; they do not confirm indexing, citations or bot visits.

Open robots.txt history to inspect earlier observations. Choose two observations on the loaded page to compare their retained text, and use the pagination controls for older observations. Repeated text can share a snapshot while retaining each crawl’s observation. A truncation notice means the retained text is incomplete. Ask agent carries the selected crawl context into a conversation about the evidence.

Work through an issue

Use Ask agent on supported evidence screens, or Work on this on the related Action. Ask for a small, reviewable correction tied to the affected page.

On a content page, such as an article or guide, the related Action also holds any content gaps on that page. The issue still appears in Issues.

For example: “Explain the confirmed issue on this page, distinguish it from unresolved checks, and prepare a fix brief for our developer.”

Check after implementation

After your team changes the site, declare the relevant Action implemented and run a later crawl when appropriate. Review the resulting checks and comparison scope. Different page coverage or incompatible conditions may limit what can be compared.