A payment acquirer doesn't only carry its own risk. Every payment it declines lands on a merchant, and every fraudulent payment it approves becomes a loss. This Latin American acquirer was scoring millions of digital transactions a year with rules too blunt to tell the two apart. A Pipl Trust Score calibrated to its own history changed that: 67% of transactions landed in a single accept rule, acceptance grew by $18.1M, and fraud dropped by $2.1M.
This acquirer processes digital payments for merchants across Latin America. Digital payments carry more risk than traditional payment methods, and the acquirer sits in the middle of every one: answerable to the merchants who rely on it to accept sales, and to the cardholders on the other side of each transaction.
Its existing fraud rules caught some fraud, but they also rejected good customers along the way. Across millions of transactions a year, the team had no consistent way to rank risk. That left two bad options: tighten the rules and decline more legitimate payments, or loosen them and let more fraud through.
The team didn't want to rebuild its stack. It wanted a score that could sit alongside its rules and show which payments had earned confidence.
For an acquirer, each goal touches two customers at once, the merchant and the cardholder. All four had to move together:
Rules can flag a pattern, but they can't weigh how signals relate to each other. At acquirer scale, 3 gaps made that cost visible.
A rule strict enough to stop fraud also turned away good customers. Each adjustment moved the problem instead of solving it.
Without a consistent risk score, the team couldn't rank transactions against each other or decide where an accept or reject line should sit.
Digital payments raise the risk baseline. For an acquirer, a false decline doesn't stay internal: it shows up in a merchant's revenue.
Pipl calibrated a custom Trust Score on the acquirer's own transaction history, so the score learned how fraud and genuine payments actually behaved in this environment. The clearest separator was distance. When a cardholder's billing location and IP location lined up, the payment was up to 2.1x more likely to be genuine. When they sat far apart, especially alongside a brand-new phone number or a burst of activity from one IP, fraud concentrated fast. With those signals scored together, the riskiest transactions surfaced on their own.
more likely genuine when billing and IP locations matched and the phone had a long history
more fraud on mismatched billing and IP locations, new phone numbers, and high IP velocity
the average risk among transactions scoring below 300
An acquirer's fraud decisions reach further than its own books. Approving a bad payment creates a loss, and declining a good one costs a merchant a sale. Pipl's calibrated score sorted transactions precisely enough that the team could stop reviewing everything and act at the edges instead: clear the payments that had earned confidence, and stop the ones that hadn't. Merchants kept more of their good sales, and fraud had fewer places to hide.
Good payments the calibrated score accepted that blunter rules would have turned away.
A single, narrow reject rule isolated a concentrated share of fraud the team could block outright.