Traffic drop investigation
Start here when Google clicks or impressions fell and you do not know why. The workflow localizes the observed drop and affected queries or pages, then produces hypotheses to verify. It does not infer cause from correlation.
Replace yourdomain.com with your GSC property URL.
Step 1: Confirm the drop and find the window
Section titled “Step 1: Confirm the drop and find the window”My organic traffic dropped recently on yourdomain.com. Diagnose what happened, find when it started, and tell me which queries or pages were hit hardest.
The assistant calls traffic_drops, analytics_anomalies and compare_search_periods to find the affected window and isolate the pattern.
Step 2: Separate traffic types
Section titled “Step 2: Separate traffic types”Was the drop in web search, Discover, or News? Break it down by search type.
The assistant calls search_type_breakdown to split clicks and impressions by surface. This narrows the investigation; it does not establish a ranking, Discover or News cause by itself.
Step 3: Identify lost queries
Section titled “Step 3: Identify lost queries”Which specific queries lost the most clicks in the affected period? Are there common topics or intent patterns across them?
The assistant calls seo_lost_queries and groups affected queries by topic to show whether the measured drop is concentrated in one content area.
Step 4: Check for technical causes
Section titled “Step 4: Check for technical causes”Were there any indexing changes on the affected pages around the same time the traffic dropped?
The assistant calls check_indexing_issues on a bounded sample of affected URLs to check for noindex, canonicalization or crawl evidence.
Step 5: Add available AI Overview context
Section titled “Step 5: Add available AI Overview context”Which query and searchAppearance rows are available around the affected topic? Show their clicks, impressions, CTR and position as additional evidence.
The assistant calls ai_overviews_impact to retrieve the available GSC rows. The tool does not build a comparison cohort or control for position, so report the rows as context rather than an explanation for the drop.
Common patterns
Section titled “Common patterns”| Symptom | Investigation area | Tool that measures it |
|---|---|---|
| All queries drop at once | Site-wide change or external search change | traffic_drops + compare_search_periods |
| Only Discover traffic drops | Discover-specific change requiring investigation | search_type_breakdown |
| CTR drops but positions hold | Query and searchAppearance context, when available | ai_overviews_impact |
| Specific topic cluster drops | Sample affected URLs for indexation evidence | check_indexing_issues |
| Gradual decline over months | Lost queries and remaining striking-distance queries | seo_lost_queries + seo_striking_distance |
Bing boundary
Section titled “Bing boundary”Bing’s current provider contract does not expose the exact adjacent windows required by traffic_drops and seo_lost_queries. Use google-bing-comparison.md for a side-by-side snapshot, and report Bing’s refusal instead of approximating the missing periods.