Google indexing monitoring
Use Project → Visibility → Google indexing to see what Google last reported about an important URL. This helps investigate a page that stopped receiving traffic, a redirected page or an unexpected canonical.
Connect Google once
- Follow Connecting Google Search Console. Enable Google Search Console API in the same Google Cloud project as your existing integration.
- Connect your Google account under Integrations. Your account must have access to the relevant Search Console property.
- Link that property under the project's Settings. Domain properties cover their subdomains; a URL-prefix property covers only its exact protocol, hostname and path prefix.
The existing read-only Search Console permission is sufficient. There is no separate API key, CrUX key or additional OAuth permission for indexing checks.
Inspect an important page
- Open Google indexing, or use Inspect or watch this URL in Page details.
- Enter the exact HTTP(S) URL on your project's domain. Keep the original path and query-string capitalization. URLs with credentials, ports or fragments are rejected.
- Choose Request Google check. Your hosting cron processes the queued request. Refresh later to see the saved result.
- Expand Review saved Google evidence. The successful-check time says when SCC received the result; Last Google crawl says when Google reported crawling the page. They are different timestamps, shown in UTC.
Google's API returns its stored index version, which can lag changes on your website. It does not run a live page test, request indexing or edit your website. For a live test or an indexing request, use Google Search Console yourself.
Read the evidence correctly
| State | What it means |
|---|---|
| Reported indexed | Google returned a PASS verdict at the saved check time. |
| Indexing concern | Google returned FAIL, PARTIAL or NEUTRAL. Read its coverage description before deciding what to fix. |
| Unchecked | No successful result is saved. This does not mean the URL is absent from Google. |
| Older than 14 days | The last successful result is older than two weeks. It can be useful history, but should not be treated as current. |
| Check failed | The latest attempt failed; the last successful evidence remains visible. A failure does not prove deindexing. |
| Unknown verdict | Google did not return a recognized verdict. SCC does not infer one. |
Compare Google-selected canonical with User-declared canonical and the intended page URL. A redirect or duplicate can legitimately use a different canonical. Combine this with Site audit and your own Search Console investigation. Missing crawl or canonical fields mean Not reported, not a confirmed defect.
Watch priority URLs weekly
Choose Watch weekly next to a saved URL. Focus on your homepage, valuable landing pages and pages affected by recent changes. You can watch up to 50 URLs per project. Cron queues up to 50 due URLs per hour across this installation; each watch becomes due seven days after its last request or attempt. A failed check gets the queue's bounded retries, then waits until its next weekly cycle.
Choose Stop weekly checks to remove a watch. Its saved evidence stays readable; a request already queued can still finish. Checks are not enabled for every page automatically.
After two successful checks report a different indexing verdict or Google-selected canonical, SCC records a change alert. The first result and failed checks create no change alert. Control personal delivery under Notifications; existing project team delivery also applies. An alert reports a change in evidence, not proof that an SEO edit caused it.
Filters, limits and troubleshooting
Search exact URL text and filter the inspection list by watchlist, reported state, unchecked, failed or stale evidence. Results are paginated in groups of 25. The list holds at most 500 saved URLs per project; other Pages URLs are not inspected automatically.
Manual requests for one URL are spaced at least ten minutes apart. Conservative local budgets allow 20 provider calls per property per minute and 1,000 per installation per UTC day, including retries. Google's quotas are shared with your other tools and can be reached sooner. See Google's URL Inspection documentation and usage limits.
- No result yet: check that cron runs and review System → Jobs. Queued work is not a successful inspection.
- Google rejected the check: confirm the API is enabled, the connected account has property access, and the exact URL lies inside that property.
- Reconnect Google: restore the connection under Integrations. Failed checks preserve previous results.
- Local budget reached: wait and request another check later; avoid repeatedly checking unchanged pages.
- Property changed: saved evidence belongs to its original connection/property. Relinking hides old evidence; request a fresh check for the new property. Old queued work is discarded.
Project editors request checks and manage watches. Authorized viewers, Ask AI and MCP clients with pages:read can read saved evidence. Ask AI can explain coverage and suggest investigations, but it cannot run inspections or alter watches. Its normal configured AI costs apply; the indexing read tool makes no Google call. Saved data remains readable when the license expires.