Many banks have automated rules in place that decline card transactions indicative of high fraud risk. In so doing, they block a large portion of fraudulent transactions. But they are also declining legitimate purchases from high-spending cardholders, who may decide to take their card business elsewhere.
A FICO study found that the highest-spending 20% of accounts represented nearly 50% of transactions and 70% of total dollars transacted. Another striking statistic: roughly 75% of the total dollar value of declined legitimate transactions—what we call false positives—came from this highest-spending group. With these declines, banks are clearly losing significant interchange revenue and causing dissatisfaction from some of their most valuable customers.
To counter this loss of revenue and build loyalty, what if a bank leveraged FICO® Falcon® Fraud Manager in the low-scoring region (likely not fraudulent) to reduce declines on legitimate transactions from these "high spender" customers? In other words, revise the customary fraud rule of something like "If Falcon score > 950, block card" to also include "If Falcon score < 100 and card is high spender, do not block card."
We tested this on a representative card portfolio, invoking a rule that for high spenders, all transactions below a given Falcon score threshold would be systematically approved (except for legitimate decline reasons, such as cvv mismatch or invalid PIN). The chart below shows the resulting incremental revenue increase from implementing this rule.
In the low score band—in other words, the low-risk transactions that previously may have been declined—we see that the added revenue from legitimate approved transactions would offset the resulting incremental fraud losses. As an example, automatically approving all high spender transactions below a Falcon score of 100 would result in about $1.8 million in incremental annual revenue; it would also improve the customer experience since no legitimate high spender transaction with a Falcon score of less than 100 would be denied.
To determine the value associated with each account, we considered tangible assets, such as profitability (e.g., revenue from interchange fees), as well as intangible ones, such as customer loyalty. Under such a rule, the majority of high spenders would experience less declines on legitimate transitions, boosting cardholder satisfaction. Of course, we had to balance this with the possible negative impact: the approval of more fraudulent transactions and the associated loss. We also factored in an estimated attrition, measured by decrease in transaction volume, after allowing fraud transactions to occur.
Of course, each institution should make its own call on how far it’s willing to tip the fraud risk/reward trade-off when approving more high spenders. But as our study shows, rethinking standard fraud strategies could open up new opportunities to improve the customer experience and build revenue.
