What was at stake.
AutoPartOcean's chargeback ratio had crept past acceptable thresholds, putting their primary processor relationship at risk. A processor termination would have stopped revenue overnight.
Bottom-line drag was substantial: representment costs, lost merchandise, and approval-rate compression as risk filters tightened to compensate.
What we did.
We started with a fraud audit — segmenting historical disputes by reason code, BIN range, basket value and geographic origin. The patterns pointed to two clusters: friendly fraud on common SKUs, and true fraud on high-ticket items shipped to fresh-account addresses.
We deployed 3DS for high-risk segments only (preserving conversion on the safe majority), built custom fraud rules with velocity caps, and stood up a dispute-representment program with evidence packs tuned to processor preferences.
Quarterly reviews tracked recovery rate by reason code so we could iterate the playbook based on what actually worked.
What changed.
Chargeback ratio down 64% over 18 months — well below threshold. Approval rate climbed 12 percentage points as risk filters could be loosened on low-risk segments.
Recovery rate on disputes that did occur stabilized at 62%, materially above the merchant's prior 27%.