Using data analytics to tackle fraud and error - new National Audit Office report published

Data analytics are a vital tool to help tackle fraud and error, which the National Audit Office (NAO) estimates cost the taxpayer between £55 billion and £81 billion in 2023-24.

• Savings so far have been modest compared to the amount potentially achievable.
• The NAO has identified ten challenges for the government to address to gainmore fraud and error savings through data analytics.

Where public bodies have implemented data analytics to tackle fraud and error, they have achieved significant returns on their investment, but savings so far have been modest compared to the amount potentially achievable, concludes a new NAO report.

Data analytics are a vital tool to make sure the right amount of money goes to the right recipient, and to find potentially incorrect transactions – this can range from basic checks that a public body has only paid a supplier once, to using emerging technology like artificial intelligence (AI) to identify risky transactions.

The Government Digital Service believes the government could save up to £6 billion a year1 by using data analytics to reduce fraud and waste, though this estimate does not account for the costs or efforts needed to achieve such savings, so should be read with some caution.

The Department of Work & Pensions and HM Revenue & Customs have been using data analytics to tackle fraud and error for a long time, with much of their work involving data matching, analysing networks, anomaly detection and predictive modelling to verify that benefit claims and tax returns match other data sources. Many other public bodies are piloting and experimenting with data analytics to make fraud and error savings.

However, there is no clear plan for how to realise the potential of data analytics to tackle fraud and error across government.

The Government Digital Service's blueprint for modern digital government sets out an ambitious vision for digital transformation, but it has not yet translated this into an implementation plan or considered that plan from the perspective of fraud and error data analytics.

The NAO's latest report examines how well-placed government is to seize the opportunity offered by old and new data analytics technologies to tackle fraud and error. It identifies ten challenges that government needs to overcome to realise more fraud and error savings through data analytics.

These challenges are: 

1. Providing cross-government leadership 

1. Scaling up and replicating projects to focus on fraud prevention 

1. Making the investment case for data analytics 

1. Making the most of central counter-fraud initiatives 

1. Building controls into existing processes and new projects 

1. Managing the key datasets 

1. Managing the data-sharing process 

1. Putting in place the right skills 

1. Optimising the staffing and algorithms to maximise the return 

1. Maintaining public trust while harnessing new capabilities 

Alongside the ten challenges highlighted in the report there are a series of recommendations, including that the Public Sector Fraud Authority (PSFA) should work with the Government Digital Service to publish a playbook on how public bodies can develop the multidisciplinary teams and capabilities to develop and deploy counter-fraud data analytics.

Alongside this, the Treasury should make the use of the National Fraud Initiative mandatory and should also agree with PSFA the criteria for where public bodies should use other centrally provided tools. 

Gareth Davies, head of the NAO, said:

“Government has the opportunity to make significantly more use of data analytics to tackle fraud and error, with potentially billions in savings available. It should learn from those who are doing this well, in both the private and the public sectors, to maximise the savings it can achieve.”  

• ENDS

Notes to editors

1. The £6 billion estimate from the Government Digital Service was based on applying the savings the Department for Work & Pensions has achieved in one example of data analytics to the estimated level of fraud and error across all of government. As such, this does not consider of the cost or effort needed to achieve the savings, or what needs to happen for such savings to be delivered. 

1. Fraud and error in the public sector generally means an incorrect amount of money has been paid out or received by government, or government has made a transaction with an incorrect or ineligible party.