How to keep Search Console inspection traffic out of product analytics
A practical measurement pattern for separating Google URL Inspection visits from genuine developer traffic without deleting the raw audit trail.
Practical takeaway
Keep the raw rows, name the exclusion, report both totals, and stop known inspection traffic before it creates a new browser identity.
The misleading signal
A live URL inspection looks like a real visit to a small analytics system. Google fetches the page, client code can run, and a new anonymous browser identifier may appear. When a site has only a few dozen daily visitors, several inspection checks can look like a sudden acquisition spike.
The pattern is recognizable after the fact: new identifiers arrive within the inspection window, each one requests a single inspected path, and none continues into a real product action. But recognizing it manually every day is not a measurement system.
The correction pattern
The safest correction is an exclusion ledger rather than deletion. Preserve the original pageview rows, store the affected anonymous daily identifiers with a named reason, and calculate corrected users and pageviews by excluding those identifiers. This keeps the measurement decision auditable and reversible.
Expose the boundary in the metrics response. Raw users, automated exclusions, corrected users, verification users, genuine actions, and verification actions should be separate fields. A public or internal dashboard can then show the number used for growth decisions without hiding how it was derived.
- Retain raw events and add a reasoned exclusion record.
- Use corrected users and genuine actions for targets.
- Display excluded counts so the adjustment stays visible.
Prevent the next false visit
Historical correction is only half the job. Reject known crawler and inspection user agents at the analytics endpoint before reading or writing the payload. On the client, stop tracking for automated user agents, webdriver sessions, and an explicit verification acquisition marker used during production checks.
The important ordering detail is to resolve the acquisition marker before creating an anonymous identifier. Otherwise the pageview request may be skipped while the browser identity still inflates the user count. A clean verification run should leave raw users, corrected users, pageviews, and product actions unchanged.