Site Health

​

Purpose

The Site Health page is a central dashboard where customers monitor how AI assistants and search engines read, understand, and cite their website. It surfaces technical, content, and discovery issues that block or weaken a site’s visibility, and lets a team track those issues through a complete lifecycle - from discovery to active work, and finally to verified resolution. By tying audit findings to real-world traffic data, the page helps marketers prioritize fixes based on estimated impact. It also provides the exact evidence, AI-generated drafts, and developer tickets needed to execute those fixes.

​

What’s on the page

​

Page header

• Title & Tabs: The page title, followed by view tabs for Overview, Issues, Results, and Details. These tabs disappear when a specific issue or page review is opened so the review has the full workspace.
• Badges: Contextual indicators like Checking now (if an audit is in progress) or Audits paused (if scheduled monitoring is disabled).
• Run audit: A button to start an on-demand full site audit.
• Options menu (⋮): A trailing menu containing Pause scheduled audits / Resume scheduled audits, Crawl scope, and Staging checks.

​

Start monitoring card

If a site has never been audited, this card replaces the main tabs. It explains what Athena checks (Site check, What AI followed, Real results) and provides a Start monitoring button to kick off the first audit.

​

Running / failed notices

Banners that appear above the report while an audit is in progress (Checking your site now, usually about a minute) or when a check fails (The last check could not finish). These banners do not hide the previous, still-valid report data.

​

Overview view

• Score tiles: Three top-level metrics with historical sparkline charts and deltas. Each opens where its number comes from: the two rates open the Issues view grouped by page, with each page’s own rates, and Can AI read your site opens Details at Check coverage:
  • AI citation rate: The share of AI answers to this site’s target questions that cite it.
  • Win rate vs top pages: When AI cites this page for a question, how much of the attention it gets compared with the top five pages.
  • Can AI read your site: Whether AI assistants can reach, read, and understand the site.
• Estimated gain banner: A text summary estimating how much the scores could lift if open issues were fixed.
• Fix first: A list of the top 5 open issues the system recommends working on first.
• Pages to work on: A list of the top 5 weakest-scoring pages (ranked by weakest win rate) alongside their open issue counts. Clicking a page opens its report.
• Results at a glance: A summary of work completed (Fixed and Verified issue counts).

​

Issues view

The full workspace for tracking problems, governed by a filter bar:

• Filter bar: Lets users toggle between List and Board layouts, Group by (Problem, Page, Status), and Search issues.
• Check kind tags: Search only marks a classic SEO check: it affects search results or visitors, with nothing to show AI assistants use it (favicon, text encoding, phone layout, mixed content, page speed and Core Web Vitals). Optional marks the classic checks that are polish rather than problems (search summary, duplicate summaries, main heading, page language, share image): an issue shows one tag, and Optional wins. Optional issues sit at the bottom of the list, never in Fix first, and their descriptions say they are optional.
• Issue band list (List layout): Rows grouped into collapsible bands (e.g., grouped by Problem or Status). Each row represents one finding (a specific problem on a specific page).
• Issue board (Board layout): A Kanban board with columns for Open, In progress, Fixed, and Verified. Each card represents a whole kind of problem (check).
• Page tree: When grouped by Page, shows a hierarchical view of the site’s URL structure. Folders open and close; the host, and any chain of single folders below it, start open, and the sections where the site branches start closed with their counts. Every page lists its issues directly under it. Clicking a page opens its report, including a page the audit found issues on but did not score (a URL nothing in the report or the issue inventory knows opens nothing). Once the reader opens or closes a folder, the open set is kept in the URL (?folders=), so it survives opening and closing an issue; closing a page’s report returns focus to its row.

​

Results view

• Result Tiles: Three metric tiles (Fixed, Verified, Still open) showing the lifecycle of all site issues.
• Your fixes: One list tracking the impact of user fixes: for each fixed issue, how often AI cited the page in the 30 days before the fix and up to 30 days since.
• Site analytics: Live traffic figures covering the last 28 days. Includes cards for Google Analytics 4 (AI visitors) and Google Search Console (Google search clicks) with sparklines and top pages/searches. Requires connected integrations.
• Pages AI cites most: A table lining up the pages AI cites most often alongside what Google sends to those same pages.

​

Details view

Deep-dive diagnostics grouped in an accordion:

• Technical checks: Cards detailing Indexing access, Page structure, Sitemap, Internal links, Page essentials, and International SEO. Also contains the AI crawler access table, International pages table, and granular Check coverage.
• Page speed: The latest PageSpeed result of each measured page, slowest first. It covers the 200 pages crawls measured most recently plus the pages the last 10 site audits measured; when crawls measured more, it says older results are left out. Tiles for pages measured, pages passing Core Web Vitals (real-visitor data only), median score, and slow pages (score under 50). Each row shows the score, LCP, INP, and CLS (lab numbers marked “lab”) and the biggest fix; expanding a row shows each metric against Google’s good and poor lines, what slows the page down with the files involved, page weight, a link to PageSpeed Insights, and Test again. Test a page picks any page from the crawl inventory and measures it on demand.
• What AI read on your site: Displays what the agent tried to answer, audience/offerings identified, and an interactive AI discovery trail graph charting the path an agent took through the site.
• Check coverage: Lists Readiness gaps (the changes most likely to help AI) and an All checks tabbed list covering Access and Understanding. The readiness score is made of six checks: AI crawlers are not blocked by bot detection, AI crawlers can reach the homepage, the homepage has readable text without JavaScript, a sitemap exists, structured data exists, and no page is too large to read whole. API, MCP, llms.txt, markdown, accessibility, brand-name, trust-page, login-wall and redirect-pattern checks were retired; older reports are scored again without them.

​

Issue review overlay (Drilldown)

When a user clicks an issue row or card, the workspace is replaced by an issue review overlay.

• All issues rail: A sidebar to navigate between issues using arrow keys or clicks, filterable by status.
• Issue details: Explains “What’s wrong” and “How to fix”, alongside a list of “Pages with this issue” ranked by AI citation impact.
• Properties sidebar: Contains a Fix section (for generating text fixes in-place), Status dropdown, Assignee selector, and a Recheck button.

​

Page review overlay (Drilldown)

When a user clicks a page in the page tree or the “Pages to work on” list, the workspace is replaced by a page review overlay.

• Page header: The URL, its citation performance, and top rivals/siblings AI cites instead.
• Issues on this page: A list of specific findings on the page.
• What’s working: What the content review found the page does well.
• Questions this page competes for: Target questions broken down by “Cited this page” and “Not cited yet”.
• Compared with the pages AI cites: The Rival comparison table.
• Properties sidebar: Page-level Fix actions, Scores (Retrievability and Citability), platform-specific scores, and page-specific Estimated gain.

​

Tables

Issue band list (Issues view, List layout)

• Status: Open/In progress/Fixed/Verified ring or checkmark glyph.
• Page or Problem title: The page path (if under a problem band) or the finding’s title.
• Assignee: Who the finding is assigned to; displayed as a dropdown menu.
• Last checked: How long ago the finding was last measured.

Issue board (Issues view, Board layout)

• Check name: The check ID in plain words (e.g., “Meta description”).
• Kind tag: Search only or Optional, when the check is one (see Check kind tags).
• Priority glyph: A 3-bar indicator for Fix now / Fix soon / Nice to have.
• Bucket badge: Category dot (technical cleanup, content restructuring, content enrichment).
• Last checked / page count: When last checked and how many pages carry this issue.
• Assignee (bulk): A menu to assign every page of this issue at once.

Page tree (Issues view, grouped by Page)

• Page: Path segment or page name. Folders expand; a page opens its report.
• AI citation rate: Retrievability score for that page.
• Win rate vs top pages: Citability score for that page.
• Issues: For a folder, the pages under it and how many issues are open; for a page, how many issues are open on it.

AI crawler access (Details > Technical checks)

• Crawler: The crawler’s name and purpose badge (Training/Retrieval/Search/Agent).
• robots.txt: Verdict (Allowed/Disallowed/Allowed by */Not listed/Unknown) and a link to disallowed pages.
• Live check: The result of fetching as that crawler (Allowed/Blocked/Challenged/Error/Not tested).

International pages (Details > Technical checks)

• Locale: Language code of the source page’s annotation.
• Page: The source page URL.
• Related target: The localized target page URL.
• What failed: The specific hreflang rule violation.

Searches that found the site (Results > GSC card)

• Search: The Google Search query.
• Clicks: Search Console clicks.
• Impressions: Search Console impressions.
• Position: Average ranking position.

Pages AI cites most (Results)

• Page: The page path.
• AI citations: Times AI answers cited the page.
• Prompts: Distinct prompts that cited the page.
• AI visits: GA4 sessions from AI referrers (if connected).
• Google clicks: Search Console clicks (if connected).
• Impressions: Search Console impressions (if connected).
• Position: Average Search Console position (if connected).

Compared with the pages AI cites (Page review overlay)

• Page: ‘This page’ row, followed by rival pages linked with their AI-answer count.
• Schema: Structured data types present.
• Tables: Number of tables.
• FAQ block: Number of FAQ headings.
• Freshness: Latest year mentioned.
• Words: Word count.

​

What you can do here

• Start monitoring: On a brand new site, clicks initialize scheduled audits and kick off the first run.
• Run audit: In the page header; starts an on-demand full site audit. Disabled while one is already running.
• Pause scheduled audits / Resume scheduled audits: Located in the top header’s options menu (⋮). Toggles the site’s weekly monitoring schedule.
• Crawl scope: Located in the options menu (⋮). Navigates to the properties page to configure crawl limits and subdomains.
• Staging checks: Located in the options menu (⋮). Navigates to the releases page for staging comparison.
• Mark all N in progress: In the Issues list, click the kebab menu (⋮) on a problem band to set every open page of that issue to “In progress”.
• Move all back to open: In the same bulk menu, resets every in-progress page of that kind back to “Open”.
• Assign all: In the bulk menu, opens a submenu to assign every unresolved page of that kind to Unassigned or a specific teammate.
• Assign this page: On any individual issue row, click the assignee face/name to assign or unassign a single finding.
• Drag card between Open and In progress: In the Board layout, drag an issue card between the Open and In progress columns. (Fixed/Verified columns do not accept drops).
• Collapse all / Expand all: Above the issues list, folds or unfolds every band in the current grouping.
• Show more pages / Show N more: At the bottom of an issue band, loads the next page of affected pages for that problem.
• Load more: At the bottom of the Issues list, loads additional issue groups when the site has a massive number of problems.
• Fix (in-place): Inside an issue/page review for titles, summaries, or structured data. Click Fix [field] to generate AI text. Shows a before/after view, a Copy new [field] button, and a Fix again button to regenerate.
• Fix with Athena: Inside a review for body text issues. Opens the Ask Athena chat pre-loaded with the page context so the agent can draft a rewrite.
• Copy developer ticket: Inside a review for code-level issues. Copies a formatted markdown ticket containing the issue, evidence, and instructions for a developer.
• Recheck: Inside an issue review properties sidebar. Starts a targeted audit to verify if the issue is gone. Depending on the issue, this runs a page-scoped, property-scoped, or full site-scoped audit.
• Status select: Inside the review properties sidebar. Manually toggle a finding between Open and In progress.
• Connect GA4 / Connect Search Console: On the Results tab, opens an in-page drawer to authenticate and link Google accounts without leaving Site Health.
• Replay route: In the Details tab’s “What AI read on your site” section, re-animates the agent’s page-by-page discovery path.
• Try again: Standard retry buttons located inside error callouts when a specific fetch (issues, fixes, analytics, dashboard) fails.

​

Data shown

• AI citation rate & Win rate vs top pages: Measured based on how often your monitored target questions cite your site’s pages in AI answers.
• Audit findings: Technical, content, and discovery problems identified by the Athena crawler reading your tracked pages and comparing them against SEO best practices.
• Agent discovery trail: The exact path an AI agent took (starting page, sitemaps, internal searches) to understand your brand and offerings.
• Site analytics: Traffic metrics (clicks, impressions, position, AI visits) pulled directly from your connected Google Analytics 4 and Google Search Console accounts, filtered to the last 28 days.
• Rival page signals: Characteristics (word count, tables, freshness) of the pages AI quotes instead of yours, pulled from live competitive profiling.

​

Common workflows

Start monitoring a new site

1. Open Site Health for a website that hasn’t been audited yet.
2. Review the explanations on the “See your site the way an AI assistant does” card.
3. Click Start monitoring to enable scheduled checks (requires write access to analytics).
4. Wait on the page as the running notice updates; the first report will build and automatically transition to the Overview tab.

Fix and verify an issue

1. From Overview’s “Fix first” list or the main Issues view, click on an issue row to open the issue review overlay.
2. Read the “What’s wrong” and “How to fix” explanations, and review the list of affected pages (ranked by their citation impact).
3. Depending on the issue type, click Fix to generate text in-place, click Fix with Athena to have the chat draft a body rewrite, or click Copy developer ticket to hand the code fix to an engineer.
4. Apply the change to your actual website.
5. Click Recheck in the sidebar to start a targeted audit. If the problem is gone, the system will automatically mark it as Verified.

Bulk-manage a whole kind of problem

1. Go to the Issues tab and ensure it is in List layout and grouped by Problem.
2. Click the kebab menu (⋮) on the right side of a problem band (labeled “Actions for every page of [issue]”).
3. Select Mark all N in progress, Move all back to open, or use the Assign all submenu to hand every page of that issue to a teammate.
4. A toast notification will confirm how many pages were successfully moved or assigned.

Move an issue on the board

1. Go to the Issues tab and switch the layout toggle from List to Board.
2. Find the issue card you want to update.
3. Click and drag the card from the Open column to the In progress column (or vice versa).
4. Alternatively, click the assignee menu directly on the card to assign every page of that issue without having to open the review.

Review site-wide results and connect analytics

1. Navigate to the Results tab.
2. Review the top tiles to see how many issues are Fixed, Verified, or Still open.
3. If your Google accounts are not connected, click Connect GA4 or Connect Search Console inside their respective cards to authenticate via an in-page drawer.
4. Scroll down to review the “Pages AI cites most” table to compare your AI citations against the organic clicks Google sends to those same pages.

​

Empty, loading, and error states

• Empty states:
  • If the site has never been audited: “Run your first audit” or “Your first audit is running”.
  • If there are no open issues: “No open issues” or “Nothing open right now” (displays the next-audit date).
  • If the page grouping has nothing scored: “No pages scored yet”.
  • If a page has no issues: “No issue is open on this page”.
  • On the Results tab: “Nothing to measure yet” or “Nothing has been fixed yet”.
  • If no page has a speed result: “No page speed results yet”, with Test a page.
• Loading states: Skeleton rows and tiles render in the exact shape of the destination view (e.g., three skeleton tiles for Overview, a skeleton toolbar/rows for Issues). Full-page transitions show a centered “Loading issue” or “Loading Site Health” status.
• Error states: Failed component reads (issues, fixes, analytics) render a Callout box containing the error reason and a Try again button. If the primary dashboard query fails entirely, a full-page “Site Health could not load” error displays with a retry button. Deep links to deleted/untracked issues show “This issue could not be loaded” with options to Try again or go Back to issues.

​

Linked from / links to

• Linked from: The left sidebar navigation (main app), legacy /seo and /site-health/tasks URLs (which automatically redirect and repair into the correct view), and back-links from sub-routes.
• Links to: Options menu links go to Crawl scope (/site-health/properties), Staging checks (/site-health/releases), and All crawled pages (/site-health/site-map). The integrations connection cards link to Google OAuth flows. The “Fix with Athena” button bridges directly into the Ask Athena chat panel.

​

Common support questions

• Why can’t I mark an issue as Fixed or Verified? The “Fixed” and “Verified” statuses represent objective measurements, not manual claims. You cannot set them by hand. The system automatically moves an issue to Verified once a later audit checks the page and confirms the problem is gone.
• Why did my Recheck take so long (or scan the whole site)? Recheck triggers different audits depending on the issue. Most checks only scan the single affected page. However, checks that compare pages against each other (like duplicate titles) require a property-scoped audit, and global checks (like robots.txt) require a full site audit.
• Why doesn’t my list show some technical issues other SEO tools report? Site Health only reports what is a problem for search engines or AI. It does not flag X (Twitter) card tags (X reuses Open Graph), image sizes (layout shift is measured directly), heading levels, link wording, title or description length (search engines rewrite them), or a missing canonical tag. Speed issues appear only when a page is in Google’s “poor” range, and the three Core Web Vitals only from real visitors’ data. A page found only through a link is not flagged for being hidden or short, since that is often on purpose. The content review judges one thing: whether the page answers the questions it is cited for. It no longer grades writing style (sentence length, facts per 100 words, the latest year written, an FAQ block on every page, answer order, tables, wording that matches the prompts, the structured data type guessed from the URL, or how much text sits outside the main content).
• What does “Challenged” mean in the AI crawler access table? “Challenged” means the site’s bot-protection service (like Cloudflare or Akamai) intercepted the audit’s request and served a captcha or interstitial. Because the audit spoofs crawler user-agents from our own cloud IPs, this challenge only proves our probe was intercepted, not necessarily that the real AI crawler is blocked.
• What is “Estimated gain”? Estimated gain (headroom) is a theoretical sum of how much your scores could improve if every open issue were fixed. It is based on typical catalog lifts for those issue types, not a guaranteed measurement or forecast for your specific site.
• Why are bulk actions disabled? Bulk actions (like “Mark all in progress” on a problem band) act on every unresolved page of that issue type. If you have an active search filter, the UI hides or disables bulk actions to prevent you from accidentally moving hidden pages that didn’t match your search.