Fraud Prevention

Leave nothing to chance – automating the fight against invoice fraud

Karim Ben-Jaafar, Senior Vice President at Quadient Accounts Payable by Beanworks argues that enterprises should first look to standardise their invoicing processes to reduce the opportunities for fraudsters.

Even with rates of fraud dropping year-on-year, it still costs UK businesses billions annually. Invoice fraud alone totals £57 million, according to UK Finance

Yet just because an issue is large, doesn’t mean it’s unsolvable. Invoice fraud may be only one type of scam businesses face, alongside cyber-attacks and other avenues, but it’s one that, using automation and artificial intelligence, we can begin to address immediately.

Karim Ben-Jaafar, Senior Vice President, at Quadient Accounts Payable by Beanworks

Understanding the enemy

Certain finance operations will always be more vulnerable to fraud. As a large-scale process that’s often still performed manually, the accounts payable (AP) process is a prime example. The exact nature of the deception may differ. It may be a case of raising an invoice for an existing customer that is then diverted to the fraudster’s own account, adding to an existing invoice then keeping the excess, or even creating an entirely fabricated business. However, the lesson is the same – the larger and more complex the system, and less oversight and control over the process there is, the easier it will be to commit fraud. 

At the same time, not every anti-fraud approach is equally viable. For instance, proactively policing employee activity to spot potential fraud may seem like a fix, but it has its limitations. First, employees may not even be driving fraud. At worst, they may be unwitting victims of an external actor. Second, it can demand prohibitive resources. And third, if innocent employees believe their employer doesn’t trust them, it will create an extremely hostile working environment. 

The simplest approach is instead to create an environment that reduces or eliminates the opportunities for deceit, and that, if the worst happens, can spot an issue and take swift action.

Standardisation and automation

The first step is to ensure control over the invoicing process. The more the process is standardised, with users only able to perform certain actions in certain ways, the fewer opportunities there will be to create or modify fraudulent invoices. For instance, invoices should only be made out to approved recipients, and should only be shared with those individuals that need to be involved in raising and approving them. This not only reduces opportunities for fraud, but standardised processes are also easier to monitor and review – making it much simpler to identify and act upon any activity that strays from the norm and might be a sign of a scam, such as an invoice with a non-sequential number. 

This process should also be as automated as possible, from raising and approving invoices to identifying anomalies. Partly, this is a matter of practicality – freeing up AP teams’ time from manual data entry and potentially tedious monitoring to focus on tasks that require more ingenuity and thought. But it’s also a matter of consistency and accuracy. An AI-driven process can automatically match invoices and vendor information, filling in every detail without the chance of human error and eliminating the risk of creating duplicate or fake documents. Similarly, an automated approval chain allows the business to only involve key stakeholders, minimising the number of people dealing with an invoice, and the risk that a single person can commit fraud. 

Likewise, an automated AI-driven process can automatically identify and flag even the smallest discrepancy in the chain, whether this means incorrect invoice data, or a new stakeholder. And every single action taken can be itemised and accurately recorded, so that the business can track and review the entire process.

Beyond fraud

While aimed at invoice fraud, this approach will inevitably provide additional benefits. It can help identify fraud regardless of the perpetrator or cause. For instance, if a fraudster is operating on the vendor end, the AP function can still identify invoices for unusual amounts, or those that don’t tally with the service provided. Similarly, a cyber-attack that affects AP functions can be detected if it creates anomalies in the process. 

Beyond this, automation can improve the entire AP process. It greatly reduces the chances of human error, accelerates payments and ultimately reduces risk. This improves efficiency and creates better relationships with vendors, which can in turn lead to improved payment terms. 

Protecting against fraud is an obligation for every business, of any size. But if done well, it can also be a way to open-up new opportunities.

Main image: