The Fraud module sits in front of POAS. It scores every click and impression for validity, removes invalid traffic from the score, and feeds the worst offenders into the exclusion lists the operator controls.
What gets scored
Every click and impression that arrives via your ad platforms is scored on three axes:
- Device and source heuristics (datacenter IPs, known bot signatures, headless browsers).
- Behavioural patterns (CTR-to-engagement ratios, retargeting loops, dwell-time anomalies).
- Outcome correlation (the placement’s historical relationship to checkout completion).
What happens to invalid traffic
Traffic flagged invalid is excluded from POAS so the score reflects only valid spend. The flagged placements and audiences flow into a recommended exclusion list, which the operator can promote into platform-side exclusions in one click or auto-apply inside the rules they own.
What you control
You set the validity threshold (strict, balanced, lenient), choose whether exclusions are notify-only or auto-applied, and own the override list for placements you know are legitimate despite a flag.
Takeaways
- Invalid traffic is stripped from POAS so the score reflects real spend.
- Exclusion lists are recommended, not imposed.
- The validity threshold is a dial; strict, balanced, and lenient are each defensible for different operator profiles.