Reducing losses without rejecting customers.
WHAT THE TYPES ARE
Stolen card details used online Account takeover of a legitimate customer Friendly fraud, where a genuine purchaser disputes Testing, where stolen numbers are validated with small transactions Refund and triangulation schemes
WHAT CARD TESTING LOOKS LIKE
Many small transactions in quick succession, mostly failing.
WHY IT MATTERS EVEN WHEN IT FAILS
It consumes fees, raises decline ratios, and attracts scheme attention.
WHAT STOPS IT
Rate limiting per address, device and card A challenge on anomalous patterns Blocking known bad sources
WHAT SIGNALS ELEVATED RISK
Mismatch between billing and delivery New account with a high-value order Several cards from one device Unusual hours and velocity Delivery to a forwarding address
WHAT TO DO WITH RISK SCORES
Set thresholds for accept, review and reject.
WHY REVIEW MATTERS
It captures value that automatic rejection loses.
WHAT TO MEASURE
Fraud rate False positive rate Revenue lost to rejection
WHY THAT LAST FIGURE
It is frequently larger than the fraud prevented.
WHAT TO NEVER DO
Optimise fraud to zero.
