Finance AI

Using advanced AI to eliminate unnecessary employee spend

by Uri Kogan January 19, 2021

Managing employee expenses is a mundane and often error-prone task that is an expensive drain on financial resources. Unfortunately, spend management becomes an even more costly shackle due to wastage or fraud in those expense claims.

The Association of Certified Fraud Examiners (ACFE) estimates that expense fraud accounts for 15% of all fraud detected. While the cost of identifying and rectifying the issue may seem much more than the potential return to the organization, this is a myth. That same research highlighted that the median duration of expense claim fraud is 24 months and a median loss of $31,000. This is an expensive problem that is not going away.

Thankfully, modern technology can help. Artificial intelligence, in particular, can automate the processing of regular expenses but can also be used to identify areas of wastage and fraud — previously unseen by manual processing.

Expense Processing Automation

The expense report auditing process is, in theory, straightforward. Auditors work through each line of each report, checking that each expense line is in line with corporate policies and has backing documentation — typically in the form of a receipt. This process has numerous challenges, including the multiple and often overlapping policies that most organizations employ, plus the sheer volume of expense reports that any large enterprise needs to work through. The colossal number of permutations that define an acceptable expense makes auditing a complex, rigorous, and specialized task. But absolutely a task that Artificial Intelligence systems can automate.

AI has numerous capabilities that enable it to automate expense processing.

  1. Expense-specific AI can understand finance documents — receipts, expense reports, missing affidavit forms, etc.
  2. AI can apply risk factors to individual expense lines and reports — enabling auto-approval for low-risk reports and routing to relevant staff for those that need manual approval.
  3. AI can achieve 100% coverage — the guaranteed application of every policy against every expense line.

These capabilities form the basis for enterprise expense automation. They increase the efficiency of the auditing process and reduce the overall cost of execution. However, they do not specifically identify wastage and fraudulent activity within reports — for that, we need to go beyond the basics.

Going Beyond the Basics

To effectively identify the nuances associated with wastage, unnecessary spending, and fraud, AI needs to behave like a human. The best auditors can sniff out a duplicate expense or a policy exception — based on gut-feel, years of experience, and often human intuition. However, those exceptional auditors are few and far between and can only hope to find a small percentage of the reports that fall into this category. To deliver this outstanding level of expense interrogation, we need to apply AI at a much higher level. Advanced expense-specific AI automation offers this via a series of interconnected capabilities:

All expenses across all dimensions

AI systems have access to vast amounts of processing power, so offer the ability to explore expense comparisons across multiple dimensions. For example, when looking for duplicate expenses, an AI can search across

  • All expenses in a single report from a single employee
  • All expenses across time from a single employee
  • All expenses across time from across all employees
  • Across all offices
  • Across all countries

This level of cross-correlation leaves no stone unturned when looking for exceptions — if finding expense exceptions is like finding needles in a haystack, AI will find all the needles.

Integrated data from other sources

Humans auditors apply common-sense and external data to expense processing all the time. For example, if Tom is working from home every Monday, why is he submitting expenses for parking on Mondays? AI systems add this context and flavor to automation by integrating data sources that enrich their view of the expense. These include internal systems such as HR records and travel portals and external data sources such as internet reviews, mapping systems, and even social media.

Learning from previous activity

Another human trait is to learn from previous mistakes — although not all humans are as good at it as others! AI platforms, however, excel at this. From an expense processing perspective, this allows an AI to “remember” that Mary has been known to add extra miles to her mileage claims in the past — this then becomes both an obvious area to check for in Mary’s reports and a known exception to check for with all other employees.

Custom Policies

Each organization has its own unique and often complex set of custom policies. The application of these multiple, detailed, custom policies across all of the dimensions described above is something that an AI can achieve far more cost- and time-effectively than human auditors.

Applying AI to Unnecessary Spend, Waste, and Fraud

We have discussed how we handle expense claims that may be fraudulent or wasteful – but what do these expense claims look like in reality? Below we detail a series of genuine scenarios that organizations are currently using AI to intercept:

Bites and Booze

    • Ensure the number of guests on a restaurant invoice matches the number on the expense claim
    • Flag an exception when a user claims a per diem expense BUT is also listed as an attendee in another person’s meal expense on the same day.

Duplicates

    • Look for duplicate expenses:

a. In different monthly reports for a single employee
b. In reports for multiple employees
c. One as a receipt and one as a corporate credit card entry

Misc

    • Track when a non-itemized receipt is rejected, and then the same expense comes back in with a Missing Receipt Affidavit (MRA).

Work from home expenses

    • Check for hotel, car rental, and meal expense claims when employees are working from home.

Travel

    • Check for breakfast expenses when included in the hotel rate.

Anti-bribery

    • Track expenses for known foreign or domestic government officials. This is required to measure cumulative figures to decrease the likelihood of abuse and to ensure anti-bribery regulations are adhered to.
    • Identify gifts/recognition items distributed to one recipient from multiple expense users.

Vendor Validity

    • Checking of the validity of vendors — especially relevant in geographies such as India where all vendors should have registered trading ID (GSTIN)

Conclusions

The automation of expense processing is the ultimate goal for many finance leaders — removing the pain of a process that consumes time and resources like no other. The application of AI can indeed deliver significant productivity and cost reductions, but the organizations that stop there are missing the real benefit of applying AI automation to expense processing.

Expense fraud is an expensive but often hidden cost to the enterprise — finding a way to manage this growing problem is of critical importance. Expense-focused AI automation is the answer to this specific problem. Deploying the equivalent of an infinite number of the best expense auditors on the planet, using AI to identify every single instance of waste, unnecessary spending, and fraud within employee expenses is the cherry on the cake of expense automation.

Uri Kogan

Vice President of Product Marketing