Plaid Says Better Approvals Beat Costly Recoveries

Fraud has traditionally been treated as a problem to investigate after money has moved. That approach is becoming harder to sustain as faster payment systems compress settlement times and reduce opportunities to reverse transactions.

For many finance organizations, the more valuable question is no longer how quickly a failed payment can be recovered, but whether the payment should have been approved in the first place.

That shift is reshaping the role of accounts receivable. Payment operations, treasury teams and technology organizations are placing greater emphasis on the quality of the authorization decision because mistakes become more expensive once settlement occurs.

Findings discussed during a recent PYMNTS interview with Shaffer Bond, payments product lead at Plaid, illustrate why. According to the latest PYMNTS Intelligence Certainty Project, 57% of firms usually detect fraud or payment nonclearance only after settlement, leaving little opportunity to recover funds once a transaction has failed.

Bond believes many organizations still devote too much attention to what happens after payment execution rather than before it.

“At the end of the day, CFOs want their settlement to go through,” he said. “What we’re seeing the vast majority of the time is that your window to recover if it doesn’t come through is limited. Making a good decision upfront is where the CFO is really going to be able to turn the tables and make sure they do get their money and they’re able to settle and get that cash flow that they need.”

That philosophy reflects the realities of faster payments. Businesses continue to compete on customer experience, and consumers have grown accustomed to transactions that complete almost immediately. Those expectations create pressure on product and engineering teams to eliminate friction, while finance executives remain accountable for limiting fraud losses and payment failures.

Bond argued that those priorities should not compete with one another because payment speed and payment quality are closely connected. Organizations must decide which transactions deserve approval before funds begin moving, rather than relying on recovery processes after the fact. The challenge is finding the balance between approving legitimate transactions and avoiding losses that become difficult to reverse once settlement is complete.

Verification Moves to the Beginning of the Payment Journey

He said organizations should examine the same information fraud investigators traditionally review after a failed payment, but apply it before the payment is approved. Customer tenure, prior transaction history, payment amounts, balances and behavioral patterns all provide signals that can strengthen payment decisions before settlement begins.

“You need to take all the things that fraud analysts would be looking at on the back end to say, ‘Can I stop this payment in flight?’ and try and get as much signal on those particular attributes upfront,” Bond said. “What you’re trying to find is what payment history somebody had or the cash flow that actually ended up with a failure. But we can review those data points at the start of the payment journey.”

Verifying who owns an account, confirming balances and satisfying basic identity requirements create the foundation upon which more advanced fraud models can operate.

Internal customer information remains essential, but Bond said this information rarely tells the complete story. Organizations also benefit from broader network intelligence that can place a new customer or unfamiliar account into a wider context. Reputation signals, device relationships and historical payment behavior observed across larger networks can help distinguish routine activity from transactions that warrant closer scrutiny.

Those capabilities ultimately depend on how effectively organizations use data.

“It’s the only thing that gets you through this problem,” Bond said, with a nod to that data. “Nothing happens in a vacuum in this world anymore. We have this technology, we have the indications of what looks good and what looks bad. It’s all very clear once you sort the data properly.”

Artificial intelligence is beginning to influence fraud strategies as well, although Bond offered a more measured assessment than much of the current market discussion. While synthetic identities and AI-generated fraud continue to evolve, he said many middle market organizations still have more immediate work to complete.

Watch the full interview with Shaffer Bond to learn more about:

• Why approval rates and return rates should be evaluated together rather than as separate operating metrics.
• How finance, engineering and product teams can balance payment speed with settlement certainty.
• What distinguishes organizations with lower fraud and nonclearance costs from those operating with greater uncertainty.