Understand traffic quality and click fraud evidence

Choose a question you can inspect: Are clicks reaching product pages? Is questionable traffic concentrated in a campaign? Do visitors from a source reach checkout?

Compare like periods and retain counts and denominators. Separate a change within a campaign from a shift toward a different mix of campaigns or sources.

These describe configuration access tiers, not a guarantee about a visitor’s intent. Missing classification and collection gaps limit conclusions. Not every bot is fraudulent advertising traffic, and a lack of purchase is not proof of automation.

Look at how visits move through relevant pages alongside their traffic evidence. A URL that looks like checkout is a suggested conversion page, not a confirmed order. An observed association does not show that traffic filtering caused an improvement.

Check whether consent, collection, classifier changes, campaign mix or small cohorts could explain a difference. Keep counterexamples: suspicious-looking visitors may still follow useful journeys.

1. Record the website, period, campaign or source, counts and comparison group.
2. Keep the observations separate from your interpretation and proposed action.
3. Record any configuration change, the previous settings and a recheck date.
4. Review platform reporting and legitimate visitor access alongside the later traffic evidence.

• Google Ads guide
• Meta Ads guide
• Read the research