Skip to main content
Index checks ask Google whether the page holding a monitored link is in its index, then keep the answer, Google’s reason and the date together. They use your Search Console connection and quota, not your link checks or credits.

Turn on index checks

  1. Open Settings › Search connections and choose Connect next to Google for the website. An admin may need to do this.
  2. Choose the Search Console property for the website. Index checks start once a property is chosen.
  3. To pause them later, use Stop index checks on the same row. Start index checks turns them back on.
Index checks cover linking pages inside a property your connected Google account can access, such as pages on sites you own. Owning the destination doesn’t give access to a publisher’s site, so a linking page on someone else’s site can’t be inspected through your property. Each website gets up to 48 index checks a day, and each page is checked once a day by default. Google allows 2,000 URL Inspection requests a day and 600 a minute for each property, and other tools you use can draw on the same quota.

Read the result

Open Links and read the In Google column. It shows by default once index checks are on, and you can add it from View. An index result says nothing about rankings, traffic or whether a backlink helped. It also never changes a link’s own present or missing state. Evidence states covers those.

How a result is decided

AgentLinkOps reads Google’s stored index record through the URL Inspection API. That API can’t test the live page or ask Google to crawl it, and a check never submits a URL for indexing.
  • Indexed needs all of these: Google’s verdict is PASS, the coverage reason is one of Submitted and indexed, Indexed, not submitted in sitemap or Indexed, indexing is allowed, Google fetched the page successfully, and Google’s chosen canonical is the URL that was inspected.
  • Not indexed means Google returned NEUTRAL or FAIL with a coverage reason outside that list. The reason stays with the result.
  • Unknown covers failed requests, missing or conflicting fields, a different Google canonical, and new positive reasons that haven’t been reviewed yet. AgentLinkOps asks for English results so the reasons match.
Every result keeps two dates. The check date is when AgentLinkOps asked Google. The last crawl date is when Google last crawled the page. A new check can return an old crawl date.

When a change counts

The first indexed result sets a baseline and sends no alert. A page counts as dropped from the index only after at least two qualifying not-indexed results, at least 30 minutes apart, and at least 24 hours after the first one. Unknown results don’t count and leave the last confirmed state in place. Repeated or out-of-order results don’t count twice. An indexed result clears any pending drop. An indexed result after a confirmed drop is recorded as reindexed. A page that was never indexed in AgentLinkOps can’t be reported as dropped. These waiting periods are AgentLinkOps rules, not Google guarantees. They show that Google kept reporting the same state, not that Google crawled the page again.

Budgets and failed attempts

Manual and scheduled index checks share the website’s daily budget. An attempt that fails or is interrupted after it starts still uses budget, and an interrupted attempt stays unsettled. A schedule is a target, so checks can run later than planned when quota or processing slots run short.

Keep the evidence

Each result keeps Google’s response and its hash, the property used, both canonical URLs and the classified reason. Google’s raw response follows your evidence retention. The dated result stays after the raw response expires. Reading a saved result doesn’t start a new inspection. For agents and scripts, use inspect_index_status for a single check, configure_index_observations for settings, and list_index_observations and list_index_events for history. The index receipts CLI exports saved evidence.

Next step

Monitor changes to follow your links and destinations alongside their index status.