Skip to content

Diagnose a traffic drop

Who it is for

An SEO, site owner or developer whose organic clicks fell and who needs a specific, evidenced answer.

The job

Name the pages that lost the clicks. Say whether the loss is a few pages or the whole site. Separate lost rankings from lost demand.

The diagnosis flow: confirm the property and two equal periods, run gsc_traffic_drop by page, split on a few pages versus the whole site, compare impressions with position to tell lost rankings from less demand from a changed results page, then inspect the worst pages.

Tools and skills

On the aeo-toolkit server. Sign in with your AEO Toolkit account. Connect Google on your project first.

  • list_gsc_sites lists your Search Console properties.
  • gsc_traffic_drop attributes the change between two periods by page or by query.
  • compare_periods compares two ranges.
  • gsc_decay finds pages slowly losing traffic.
  • gsc_url_inspect checks Google’s index status for up to 25 URLs per call. It is read-only.
  • list_ga4_properties and ga4_run_report let you check whether analytics agrees.

The method is in the seo-traffic-drop skill. Related skills: seo-content-decay and seo-cannibalization.

Steps

  1. Call list_gsc_sites and confirm the property. https://example.com/ and sc-domain:example.com are different properties.
  2. Pick two equal-length periods that align on weekdays. Comparing 30 days to 28 manufactures a drop.
  3. Run gsc_traffic_drop with dimension: 'page'. Read the clicks lost and gained before the list.
  4. Look at each page’s share of the decline. A few pages carrying most of it points at those pages. A thin spread points at the whole site.
  5. For the top pages, compare the change in impressions with the change in position:
    • impressions down and position worse: lost rankings
    • impressions down and position flat: less demand, or a changed results page
    • impressions flat, clicks down: something on the results page is absorbing clicks
  6. Re-run with dimension: 'query' for the affected pages. One topic means a content or intent problem. Scattered queries suggest a technical or site-wide cause.
  7. Run gsc_url_inspect on the worst pages to rule out an index problem.
  8. Ask what changed on your side: a release, a migration, a robots.txt edit.

What you get

A statement like “Clicks fell 2,400 between the two periods. Four pages account for 78% of that. All four lost rank, and impressions fell with them.” Then a next step per group of pages.

Limits

Search Console shows what happened, not why. It does not know about your deploys, your competitors or algorithm updates. State a cause only when the numbers and a known event line up, and call it a hypothesis.

AEO Toolkit is an open-source instrument by Advance Labs.