While the growing use of expense management solutions has made it easier and faster for organisations to automate, monitor and manage T&E claims, it is also helping to reduce available opportunities for would-be fiddlers and fraudsters.
But given the growing availability of advanced data analysis tools and algorithms, some believe that expense management solutions must evolve to include ever smarter fraud detection capabilities.
Not all suppliers in the sector are convinced, however.
Cloud-based booking and travel expense management provider KDS conducted user surveys and found that around 25% of people admit to embellishing their expenses or bumping up the mileage on car travel claims.
The real figure is probably higher. But vice-president Kate Roe says that while a good expense management tool can help prevent such abuses, it’s the operational savings that are more significant.
“One of our clients, a large multinational in the telecoms sector, has around 100,000 people regularly submitting expense claims,” she says.
“Before the company implemented expense management software, it would take each claimant around an hour to fill in their monthly claim, then another hour or two for the business to process it. That worked out at around 150,000 days the company was spending every year just on processing expense claims.
“This represents a much bigger cost to business than expense fraud. I’d question how useful it would be to apply fraud detection analysis when you’re only likely to find a little drop of savings in an ocean of waste.”
As Roe says, expense management tools are already reducing the opportunities for fiddling. They increasingly hook directly into the booking systems of approved accommodation or travel providers, eliminating the requirement for paper receipts or manual claims filing.
Where paper receipts are the only option, employees can typically just take a photo of them using a smartphone. They will be automatically read by optical character recognition (OCR) software, along with time and location data, and sent to the system.
Some handwritten or faded thermal receipts may need to be read and approved manually, but today’s OCR technology typically has an accuracy level of around 95%. And if it’s company policy to have to take a photo of receipts as soon as they’re given to you, this eliminates the ‘screwed-up in my pocket for three weeks’ illegibility problem.
The tools also allow for the automatic checking of expense claims against corporate policy and compliance rules, with approvals required from managers for any claims outside the rules or permitted spending limits.
Anthony Sherwin, chief client officer at cloud-based expense management provider Software Europe, says: “There are many ways of reducing the opportunity for fraud by having a clear expense policy and the correct flags in place. For example, an expense management system would prevent situations like manipulating a restaurant bill to include a large drinks tab in the food section of the bill, or claiming food over a certain cost.
“One of our customers had a couple of employees who were romantically involved and it was discovered that they were both arranging business meetings that required an overnight stay. They would put in separate claims for hotel rooms, when in fact they were sharing a room and pocketing the cost of the unused room.
“Reports can also flag suspect expense claims like ‘double dipping’ – when someone claims for the same item twice – or individuals who are repetitively not submitting receipts with their claims. Other useful reports can be based on particular employees’ expenditure and whether this suitably equates to their position in the company. These comparison checks and data analyses are certainly a very useful tool for our customers in spotting fraudulent patterns.”
Chris Baker, UK head of enterprise at expense management solution provider Concur (now part of SAP), agrees. “Prevention is always better than detection,” he says. “That’s where a lot of our work is focused – developing the ability of the product to prevent certain types of fraud.
“For example, the system will detect ‘buddy meals’ – where two people get a receipt for the same meal and both claim for it on expenses – by looking for the same names on expense reports from different people. Basic, rules-based reporting does catch people out, but you need a good, clear policy and some kind of audit functionality to check claims against the organisation’s policy and flag things up for investigation if they don’t look quite right.”
Baker says more advanced data analytics comes into play when Concur is deciding on what reporting capabilities and rules-based analysis to include in the solution. “We’re processing millions of expense claims per quarter across many customers. That gives us a huge amount of data that we can usefully interrogate, uncovering patterns in the aggregated claims which help us continually improve the product. Our analyses uncover a lot of interesting statistics – such as just 5% of employees being responsible for 85% of fraud.”
Nonetheless, Baker also believes more advanced analytic capabilities will increasingly be embedded in (or easily connected to) the expense management systems themselves. Already, for example, third-party expense analytics apps for Concur, such as Oversight, are beginning to appear, and are likely to become more advanced.
“The future of predictive learning lies in bringing together multiple systems,” says Baker. “In terms of detecting fraud, you need an expense management system that can pull in data from, for example, your CRM [customer relationship management] and accounting systems, correlating data and looking for patterns across all of them. We don’t know how big a problem expense fraud really is, but as paper receipts become a thing of the past, people will undoubtedly find clever new ways to defraud the system that will require ever more advanced analytic techniques to detect.”
Read more about expense management
- Gartner looks at the benefits of expense management tools and the circumstances that make software as a service preferable.
- This analyst paper from Quocirca looks at the benefits that every organisation should be looking to get from a travel and expenses management system.
Indeed, some large corporates and financial companies are already using such techniques. The big strategic consultancies and analytics specialists say they are seeing increased demand among blue chip clients to include expense data in broader analyses looking across multiple systems for indications of fraudulent behaviour and non-compliance.
Stephen Shelton, a director in the forensics team at PwC, says: “Expenses are a big cost bucket. And since analytics is becoming more accepted as a business tool, expenses seems an obvious area to apply it. There are forms of expense fraud that are difficult to spot in the rules-based workflow process typically used by expense management tools. Double dipping is one. Persistent claims below certain thresholds is another.
“Each claim might comply with the rules, but the broader pattern of behaviour could indicate something is amiss. For example, one common approach among our customers is to look at the history of expense claims, both for individuals and departments. Over time, certain people or groups may have a very different claims profile to others. There might be legitimate reasons for that, but it gives you a basis for further investigation.”
Shelton says PwC uses techniques such as clustering and predictive modelling to detect anomalous patterns in data. It’s often done as a one-off exercise for corporate clients looking to uncover potential anomalies in their expense claims history and other data. Typically, this would cost an organisation in the region of £30,000 to £50,000, he says.
Business intelligence and analytics company SAS is also seeing a rise in demand for more advanced expense analytics. Sundeep Tengur, a business solutions specialist at the company who has worked extensively with blue chip banking and finance companies, says: “A lot of companies are moving towards holistic surveillance. Since putting in such advanced analytics is a significant investment, it is not yet economical to focus solely on expense fraud. Rather, companies are looking to detect fraud across the board.
“Most of them are currently at the stage of using very simple exploratory tools, looking for anomalies in data they’ve never previously analysed. But the next stage is to apply more intelligence to mitigate risk and define a process to tackle anomalies.”
By applying advanced analytics to expense data in combination with CRM and finance system data, Tengur says it’s easier for companies to detect things such as collusion among employees, insiders who may be part of an external fraud ring, bribery and corruption, even the illicit funding of terrorism.
That such advanced analytics tools will eventually find their way into (or be able to connect easily to) off-the-shelf expense management systems seems a growing inevitability as the technology becomes ever more mature.
The challenge in future may well be more cultural than technical – ensuring that employees are reimbursed quickly for any expenses they genuinely incur on behalf of the organisation, without unnecessary red tape, while also avoiding the danger of making people feel as if they are being spied on or treated with suspicion by a distrustful, penny-pinching employer.