What went wrong, and what we learned

by Joseph K. Clark

“No one is going to ask about that – only the geek on Computer Weekly.”

Iain Duncan Smith was clearly angry with me for puncturing his wishful thinking over the reception of yet another announcement of delays to Universal Credit. It was the early autumn of 2014, and Iain was secretary of state at the Department for Work and Pensions (DWP), and I was his minister for welfare reform, responsible for bringing the project in.

We had set out on the project to reform the UK welfare system in 2010. The existing system was a mess – a bundle of often contradictory benefits that categorized and trapped people in the design and poverty over the years. With Universal Credit (UC), we wanted to sweep this all away and replace it with a system that allowed people the flexibility to change their situation in a low-risk way.


The trauma of UC’s introduction proved far beyond anything we had envisaged. In the first five years, we worked our way through no fewer than six project managers and six senior responsible owners, with several collapses and even one death under enormous pressure.

Computer Weekly kept a close tally of our travails throughout this period, fed – we suspected – by hostile informants in the Cabinet Office. It was not surprising that Iain came to dread the following revelation from this source.

IT capabilities

Why was it all so complicated?

At the heart of UC’s troubles lay the capability of the department in IT. It became evident that the commissioning out of IT by government departments a decade and earlier had effectively removed any direct knowledge of how to build systems or even monitor that the contractors were building those systems properly.

This did not matter for more minor, discrete pieces of software, where the DWP had developed a good track record. Indeed, Joe Harley, government CIO at the outset of the project, reassured me on the prospects for UC with the example of the successful recent implementation of the new Employment and Support Allowance.

“If departments do not rise to the challenge, there will be plenty more horror stories for Computer Weekly to lay bare.”

David Freud, former minister for welfare reform

But this lack of competition meant that the department did not recognize the scale of the system we wanted to build. We were still calling it “an IT development of moderate scale” in the whitepaper of 2010 and beyond.

There were two other errors at the outset. DWP did not realize the security issues involved in offering claimants an interactive system. Too late, it attempted to address the challenge by retrofitting security into a legacy-style system, discovering the outcome to be hopelessly complicated for users.

At the same time, the first attempt to use new agile development techniques fell apart because the approach used – building up individual storylines – was highly inefficient. It was also severely undermined by its incompatibility with the central government clearance timetable, leading to subsequent – misplaced – accusations of lack of financial control from the National Audit Office.

The more rapid development schedule promised by agile also allowed an accelerated timetable. The failure to achieve this schedule was the direct source of much of the public criticism of the project.

Five-way tussle

As the project’s problems escalated, a desperate five-way tussle developed within the government, which played out through 2013. The Cabinet Office exerted direct control through the Major Projects Authority and its oversight of contracts. While aggressively forcing benefit cuts on the department, the Treasury nevertheless proved supportive of its UC plans at this time, thanks to the Liberal Democrat chief secretary to the Treasury, Danny Alexander. The prime minister, David Cameron, was nervously supportive, and the Liberal Democrat deputy prime minister Nick Clegg much more positive.

The Cabinet Office was also ambitious to achieve its “digital by default” agenda. It was explicit in wanting the initial system we had built scrapped and replaced by a new digital system built from the ground up. It put considerable resources into this redesign, led by the formidable Tom Loosemore, the Government Digital Service (GDS). However, while GDS understood how IT might work, it also suffered from ignorance – the polar opposite of the departments. It desperately under-estimated the complexity of a welfare system dispensing around £100bn a year. Accordingly, the initial timetable for the digital system started falling back at a rapid rate.

The outcome of the tussles was the roll-out of two distinct systems, dubbed the “twin-track”. The first system was shaped like a standard legacy benefit, albeit one in which claimants could make an application online. After that, the interaction with DWP was through the traditional post and telephone communication. Nevertheless, the department learned how the vital operational elements should function through this system and the behavioral response from claimants to the new arrangements.

Those findings helped shape the “full service” digital system, developed and rolled out afterward. It also maintained a sense of momentum at a critical time. Clearly, there were extra costs involved in this twin-track approach, although they were offset by the gains involved in putting people on the system earlier.

As is often ineffective responses to the crisis, the twin-track approach may well have been the best way to introduce a complex new system. The early version acted as a pathfinder to inform the structure of the full-service offering.

Lessons learned

The subsequent response of the department to this baptism of fire has seen a wholesale restructuring of its IT capability, pulling development and maintenance of systems in-house and integrating its services more closely with operations. DWP has built a genuinely agile capacity in action. It has also given more authority to the operational teams so that the detailed structuring of UC reflects the behavioral responses from the claimant base.

These changes represent a challenge to other government departments considering significant developments. Purely in IT, the lessons learned imply bringing development capability back in-house, building big integrated teams to adopt agile technology, and reversing a policy-led culture to one reliant on operational feedback.   If departments do not rise to the challenge, there will be plenty more horror stories for Computer Weekly to lay bare.

David Freud was one of the architects of Universal Credit and was minister for welfare reform in the coalition government from 2010 to 2015. His book about the implementation of Universal Credit, Clashing Agendas: Inside the Welfare Trap, was published in June 2021.

Related Posts

Leave a Comment