Without automated systems, ELFS couldn’t operate. Processing around 1.3 million invoices a year for 14 different NHS trusts, the business services provider handles an enormous amount of data every day. Errors would be costly, not to mention damaging to their reputation.
Melanie Roskell, Procure to Pay Manager at ELFS Shared Services said: “We couldn’t handle the amount of work we do manually, so we automate as much as we can. We’re constantly looking to spot gaps in our provision and learn where we can improve.
NHS clients expect ELFS to run an invoice duplicate checking service. Not to have a system in place would be a “big red flag” said Melanie.
“We have to make sure that our systems and controls are as strong as they can be, to provide reassurance and a quality service to our clients,” she said.
Unable to split duplicates by client
Up until 2014, manual invoice checking methods had been fraught with problems.
“We were a much smaller business back then and still relied on many paper systems. Always prone to human error, it was possible to enter an invoice incorrectly and we had no method to pick that up. A supplier or client would then highlight the issue to us.
We worked with a company to carry out quarterly audits on our account processing. They charged commission on money recovered, so we were motivated to reduce potential errors in the first place,” said Melanie.
ELFS embraced an automated invoice checking process as the workload grew. But on reviewing this solution, ahead of the contract renewal, they highlighted a couple of frustrations. “They wanted us to move to their new platform. That meant we’d have to change many of our processes. For example, we check for duplicates daily and download the potential issues to Excel, splitting them out for different teams to review. They then feed information back to the team who action duplicates on the system. On the provider’s new platform, this was only possible at significant additional cost. It would have restricted us to reviewing all issues on-screen only.
We’d also been experiencing loads of false positives, despite the team tweaking the risk level countless times to try and manage this. That meant we were having to manually discount many ‘errors’ the platform surfaced,” explained Melanie.
ELFS discovered two providers of AI-powered invoice checking: Xelix, and a US-based provider (who was quickly discounted due to an uncompetitive price).
Due diligence before switching
Moving to a different system always carries risk. And Melanie acknowledged ELFS’ preference was to stick with what was known. So, the potential move to Xelix required in-depth consideration and a strong business case.
“There was much bias towards our (at the time) current provider because, fundamentally, it did the job. And moving to a new product is always risky. But Xelix really worked hard with us. First, they provided a test system, free of charge. We put our real data into this, comparing results with the existing platform. Xelix was clearly surfacing more accurate results. This helped to justify the move to our board.
Xelix was ticking all the boxes and provided more flexibility. So, we asked our team to have a look. Whilst they used the existing system daily, they preferred the look and feel of Xelix, finding it much more user-friendly.
Finally, we got an external accountant to review Xelix. We did so much due diligence because we’re never keen to move away from ‘tried and tested’.
The decision process took a while but the team at Xelix remained helpful and patient. They were very willing to cooperate with us, so we decided to switch. Looking back now, I know it was definitely the right decision,” said Melanie.
Preventing millions in invoice errors before payment
Melanie described the move to Xelix as “one of the smoothest projects I've worked on for quite some time.” She said Xelix had promised a quick and easy transition and “they delivered.”
The only glitch was internal training, so Xelix put on extra training sessions to “mop up the gaps” that Melanie identified.
As for the issues that prompted the switch: “With Xelix, we could split our data and download it straight away. There’s also an option for further cross checks as we develop our use of the system. As for false positives, Xelix’s AI has a smarter way of eliminating them. Whilst they initially appeared, the tool’s machine learning element identified patterns and quickly reduced false positives, increasing accuracy.”
Melanie has also been impressed with the ongoing support provided by Xelix: “We have regular reviews with a team of named people at Xelix. I know I could email or phone anyone and they’d come back to us straightaway.”
ELFS works to a regular daily process with Xelix. There’s a strict 6pm cut-o for daily postings. Xelix then checks the data overnight, surfacing potential issues in the morning before any new postings. And if SFTP fails, there’s a back-up procedure in place, says Melanie.
Whilst it’s clear having Xelix in place reduces administration time, quantifying that benefit, Melanie said, was hard. However, she confirmed that £7 million of duplicate payments (a mix of historic payments and due payments) were identified by Xelix in the first month of operation. Since then, Xelix has prevented a further £7.3 million of duplicate payments leaving the business.
Tackling duplicate vendors, fraud, and reporting
Now duplicate invoice checking is established with Xelix, Melanie has earmarked more automated processes to develop within the system.
“We have problems with duplicate vendors, perhaps because of a merger or a site name change. So, we’re going to look at the Vendors module in Xelix next. And I understand Xelix can also help us with checks on supplier bank accounts too. At the moment we do this manually. Automating these processes will be a huge benefit to us.
The fraud reports look really useful too, so I’ll be looking to develop this area soon. Fraud is massive in the NHS and we’ve already got some checks in place. But there are additional features in Xelix I want to include. For example, Xelix can highlight an outlying invoice number that isn’t in the usual sequence. We’ve not been able to do that before. To me, that’s a big flag and something we can use Xelix for quite quickly.
We’ve gone through the reporting side of the system and we’re keen to harness this soon. Though, we already have dashboards for each client, so we’ll be looking to enhance these with data from Xelix. It will be great to demonstrate the strength of protection we’re providing by using Xelix.
There are plenty of areas within Xelix for ELFS to grow into.”
