Firms auditing expenses through manual processes and outdated technologies risk missing the chance to flag erroneous or fraudulent submissions from vendors and employees.
Deploying technology — especially artificial intelligence (AI) — can streamline the auditing process and boost cost savings significantly.
The data was based on expense report, contract and invoice audit data from billions of transactions across hundreds of enterprise customers from April 1, 2019, through June 30, 2019, AppZen said.
Firms that do leverage AI to dig deeper (through deep learning and natural language processing) into their spending trends are, in fact, able to audit nearly all invoices and expenses, the report found.
Moreover, in terms of the larger picture, more than 96 percent of enterprise spend beyond that of payroll is related to accounts payable; 3.7 percent of the remainder is T&E related.
AI, according to the findings, can flag 8.7 percent of expenses as high risk, tied to duplicative spending or as being unauthorized. Similarly, about 4 percent of invoices are flagged as high risk
In an interview, AppZen Chief Marketing Officer Jamie Barnett noted that “although AI technology is relatively new, it’s triggering more interest because it’s able to complete the manual review that no one could (or wanted to do) previously.”
That comes against a backdrop where manual auditing has been the status quo, and where they may not have been aware of tech-driven solutions.
On average, the firms queried processed 200,000 expenses across the past quarter, and with more than 60,300 invoices processed per month, as many as 7 percent of those invoices were considered high risk due to inflated pricing (compared by AI to external market pricing data) or information mismatches.
AppZen also found that across every 10,000 expenses submitted, two contain regulatory violations and for every 10,000 invoices, one reveals a regulatory violation. Shedding some light on those violations, said Barnett, “we’ve seen it all, from gifts to sanctioned organizations, payments to foreign officials or politically-exposed people, to meals with healthcare professionals. Many would be difficult, if not impossible, to find without the help of AI.”
Asked about surprising findings amid the research, Barnett said that the data shows that in some cases AI has identified employees making purchases on their credit card, creating an expense report for reimbursement and submitting invoices to be paid by their accounts payable department.
In one example, Barnett told PYMNTS: “One example we found was an expense that was $45,000 for an event at a hotel. The employee was paid out, and so was the invoice. Conventional wisdom might tell you that this would have been caught, but in this case, it was approved, which goes to show how little oversight there is between expense and AP systems.”
Oversight becomes especially important as businesses are becoming ever more global in scope. Employees are traveling more than ever and are submitting expenses from all over the world, she said. AI, she added, can enforce regulatory compliance and verify receipts across dozens of languages.
Embracing alternative solutions can pay off, she added, pointing to the fact that 85 percent of companies can automatically approve low risk spend, saving hours that can be redeployed toward other business functions and strategy.