A fraud rate can look precise while saying very little about business impact. The useful question is not how much traffic looked unusual, but which anomalies changed acquisition cost, conversion quality, and downstream revenue.
Start with evidence that can be explained to a media buyer: repeated device behavior, impossible interaction timing, source-level conversion gaps, and outcomes that diverge after the click. Each signal should answer what happened, how confident the system is, and what decision it supports.
Then compare cohorts. A source with a higher anomaly rate may still be profitable, while a cleaner-looking source may produce low-quality customers. JuicyTraffic keeps these two ideas separate: traffic risk describes the evidence, while outcome analysis describes the commercial effect.
The operating habit is simple: investigate material changes, validate them against outcomes, and only then adjust spend or rules. That creates an audit trail instead of an opaque blocklist.