Digital and data challenges could hamper DWP’s fraud strategy, NAO warns

Public spending watchdog positive about department's progress in bringing multi-billion-pound benefit overpayments down
DWP's Caxton House HQ. Photo: Google Maps

By Jim Dunton

22 Oct 2025

Digital and data challenges such as “legacy” IT and a lack of common standards could hamper the Department for Work and Pensions’ ability to successfully implement its fraud and error strategy, the National Audit Office has warned.  

It said DWP’s IT systems are not fully integrated and do not allow staff to view all the information that the department holds about claimants, making it less likely that incorrect payments will be prevented or detected.  

The public spending watchdog said DWP estimated that it had overpaid £73m in benefits during 2024-25 because it had failed to take individual claimants’ full benefits income into account when calculating their entitlement. It added that the ability for claimants’ addresses to be entered into DWP’s systems using multiple formats also increased the difficulty of matching different records held for the same claimant.  

The NAO’s data concerns come in an otherwise positive report on DWP’s progress in reducing multi-billion-pound benefits overpayments, which rocketed after controls were loosened at the outset of the Covid-19 pandemic. 

Back in July, Civil Service World reported data from DWP’s annual report and accounts, which showed benefits overpayments had reduced from £9.7bn in 2023-24 to £9.5bn in 2024-25 – the latter estimate including £190m in state pension overpayments. 

In particular, the overpayment rate for Universal Credit dropped from 12.4% in 2023-24 to 9.7% in 2024-25. The NAO said interventions including the use of machine learning and the Targeted Case Review programme had resulted in savings estimated at £4.5bn since 2022 because of counter-fraud work. 

Nevertheless, the NAO said DWP has more work to do if it is to make good on its ambition of returning benefits overpayment rates to pre-pandemic levels. Figures in the report, published today, say the all-benefit overpayment rate is estimated at 3.3% in 2024-25, but would need to return to 3.1% – as measured by a “cross-welfare” rate that also includes Tax Credit overpayments – to get back to 2019-20 levels. Figures in the Spring Statement forecast that it could take until 2028-29 for that 3.1% figure to be accomplished.  

Successive governments have allocated DWP £6.7bn of dedicated funding for fraud and error activity over the nine years from 2020-21 to 2028-29, including initiatives such as Targeted Case Review – which has thousands of staff carrying out checks.

Among its recommendations, the NAO said DWP should look beyond its ambition to reduce the overpayment rate to pre-pandemic levels and focus on achieving a level that represents a “cost-effective control environment”.  

It added that the department should also look at which other overpayment areas could benefit from the kinds of detection and prevention activities that are being focused on Universal Credit. The NAO suggested Pension Credit, which had an overpayment rate of 10.3% in 2024-25 as one candidate.  

In relation to its data-quality concerns, the NAO called on DWP to put its data about benefit claimants into a common format and ensure the department is in line with cross-government data standards. 

It also said DWP should build on its existing use of data analytics to explore how emerging technologies may help to detect and prevent fraud and error. 

NAO head Gareth Davies said consolidating work to date and improving processes and controls to stop overpayments before they occur would be the best way to secure value for money in relation to fraud and error. 

“The DWP has made real progress in tackling the levels of benefit overpayments due to fraud and error, but there is still a way to go,” he said. “With the increase in funding and the greater focus on prevention, the next few years will be key to its success in addressing this long-standing issue. 

“The government should carefully consider the challenges and the recommendations outlined in today’s report if it is to build on its progress so far.” 

Public Accounts Committee chair Sir Geoffrey Clifton-Brown said service-modernisation and removing data barriers would be key to reducing overpayments further.  

“It is encouraging that machine learning is being deployed as a valuable tool in addressing fraud and error,” he said. 

“DWP must use the dedicated counter-fraud and error funding it has been awarded to go further in its approach, continue to reduce the proportion of benefit expenditure that is overpaid and ultimately protect the taxpayer.” 

Transformation minister Andrew Western said DWP is proud of the progress it has made in reducing fraud and error in the benefits system.  

“The overall rates have dropped for the first time in two years, and we will go further with our fraud bill, which is part of wider plans that will save £9.6bn by 2030,” he said.  

“As the head of the NAO has recognised, we’re rightly making use of technology, and our one machine learning model has made us significantly more effective. There are numerous safeguards in place, and final decisions in relation to fraud and error are always made by a human.” 

Read the most recent articles written by Jim Dunton - Six sentenced over £20m HMRC fraud

Share this page