Case study: pre-broker fraud scoring lifted accepted-lead rate to 94%

Case study: pre-broker fraud scoring lifted accepted-lead rate to 94%

What it is: A lead seller's broker kept rejecting batches for fraud-flagged traffic (bots, recycled data). Adding a device-and-behavior fraud score before delivery cleaned the pipe.

Best for: Lead sellers whose broker scores incoming leads and penalizes bad batches.

What was done:
— Scored leads on device fingerprint, form-fill speed, and IP reputation
— Set a rejection threshold tuned against the broker's own flagging history
— Held flagged leads for manual review rather than auto-dropping

Outcome:
— Broker-accepted rate rose from 79% to 94% over 4 weeks
— Form-fill speed under 4 seconds caught most bot submissions
— Account standing improved enough to unlock a higher payout tier

Pros:
— Behavioral signals (fill speed) caught bots that IP checks alone missed
— Manual-review queue avoided dropping borderline-real leads

Cons:
— Aggressive thresholds reject fast-but-real mobile users
— Fraudsters adapt fill timing once they learn the rule

Who should skip this: Sellers whose broker accepts on volume and doesn't score quality.

Verdict: Score fraud before delivery.