Automation and AI Move Deeper Into Payment Exceptions

In other cases requiring manual oversight, payments don’t match the invoices, ledgers or bank records surrounding them.

The August PYMNTS Intelligence report “Payment Protection: Why Firms Still Aren’t Real-Time Ready” found that 70% of surveyed firms plan within 12 months to adopt or expand automation that matches and clears payments against invoices, ledgers and bank statements. Fifty-nine percent plan to add or expand artificial intelligence (AI)-based fraud detection.

Discrepancy analysis adds another task to payments oversight. Finastra’s new Repair Recommendations capability uses AI to identify payment discrepancies, analyze their underlying causes and recommend corrections, with employees reviewing and approving the recommended action.

Payment matching, fraud detection and discrepancy repair address different parts of the same operating workload.

More Payment Work Is Moving Into Software

The May PYMNTS Intelligence report “Early Detection: Why Top-Performing Firms Focus on Fraud Before It Starts” found that 88% of 60 surveyed U.S. middle-market firms experienced at least one accounts receivable integrity issue during the previous 12 months.

Seven in 10 experienced ACH returns involving invalid or closed accounts or customer input errors that prevented clearing. Sixty-three percent experienced reversals or disputes after delivery, and 47% reported account identity and legitimacy issues.

We’d love to be your preferred source for news.

Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!

Add as Preferred Source

An ACH return carries information about why a transaction was returned. A reversal or dispute requires attention after the original transaction has progressed further. Account identity issues put verification into the workflow. Reconciliation deals with payments that have to be matched against the financial records associated with them.

Companies are spending across each of these functions. Seventy percent of firms plan near-term investment in automated payment matching, 65% plan to adopt or expand identity verification and 59% plan the same for AI-based fraud detection and secure bank connectivity.

AI Moves From Detection Into Repair

AI-based fraud detection identifies transactions that meet criteria for further attention. AI-based discrepancy analysis applies the technology after a processing issue has already been identified. We found that 57% of surveyed firms typically discover fraud or payment nonclearance after they have treated a payment as completed, through a return, dispute or contact from the receiving bank. Seventeen percent usually identify the issue before initiation and 13% during authorization.

Across the 60 surveyed companies, fraud and nonclearance costs averaged 31 basis points of revenue, which indicates that the dollars add up when there’s a payment hiccup.

Among firms that usually detected fraud or nonclearance before settlement, 81% used instant bank account verification, compared with 47% of firms that typically detected issues afterward. Open banking-based account ownership verification was used by 76% of firms detecting issues before settlement and 35% of firms detecting them afterward.

Exception Work Persists After Processing Is Automated

The same pattern appears in accounts payable, where high adoption of automation has not eliminated manual invoice work. PYMNTS reported in August that 89% of organizations use at least some AP automation, yet 67% still spend five days or more each month processing invoices. Employees continue to copy information between systems, chase approvals, review invoices and resolve exceptions.

Exception automation also changes what payments teams need from their systems. Once software takes on more matching, detection and diagnosis, employees need the information behind an alert or recommended action to decide what happens next.