Clashing agendas, by David Freud - review: How Universal Credit got there in the end

It was nearly 10 years ago, in October 2011, when Computer Weekly first reported on concerns about the IT project behind Universal Credit. That was five months before the new Coalition government’s flagship welfare reform policy had even received Royal Assent to pass into law.

Over the next few years, we followed the growing problems of the controversial policy, often becoming the first publication to reveal the latest details of the highest-profile digital programme in government at that time – one that was labelled as yet another example of costly IT failure.

This was a pivotal time in the history of government IT. The Government Digital Service (GDS) had recently been created, promising to sweep away old behaviours and past habits; to end the era of expensive outsourcing deals signed with the oligarchy of industry giants, and replace it with lower-cost, agile, nimble projects that rapidly delivered online public services with a promise to radically improve the relationship between citizen and state.

Universal Credit was to be the exemplar for this approach – proof that the geeks could educate crusty Whitehall mandarins in 21st century techniques in even this most complex of policy reforms. In digital government terms, it was to be the triumph of the new against the old; the modern against the traditional. It didn’t turn out that way – but perhaps we can now reflect that the Universal Credit debacle did after all lead to a hugely beneficial period of change in government IT.

The architect

At the heart of Universal Credit in every sense of the word was David Freud – better known at that time as Lord Freud, minister for welfare reform in the Coalition. Freud was the original architect of the proposals to overhaul and simplify the UK benefits system that became Universal Credit, first as an advisor to Labour under Tony Blair and Gordon Brown, then ennobled by David Cameron to lead its implementation.

Freud has published a memoir detailing the challenges of shepherding such a transformational policy – one that claimed the scalp of a secretary of state and senior civil servants along the way. Computer Weekly’s reporting was often dismissed by the Department for Work and Pensions (DWP) at the time through persistent denial and obfuscation, so I have to admit to a certain sense of gratification when reading Freud’s entertaining book, Clashing Agendas: Inside the Welfare Trap, to learn that our reporting had an impact at the highest levels.

“No one is going to ask about that, only the geek on Computer Weekly,” Freud recounts DWP secretary of state Iain Duncan Smith saying, in response to likely problems with the migration of data from legacy benefits to Universal Credit.

Duncan Smith was right – our subsequent story revealed “audible gasps” from DWP staff on being told they would need to transfer 200,000 benefit claimants to the new IT system every month for over a year, a migration light years beyond any such exercise ever attempted by government.

But for any observers of digital government, it’s the political, technical and cultural clashes between DWP and GDS detailed by Freud that are particularly fascinating.


It’s astonishing to read that the 2010 policy paper on welfare reform, produced for the newly formed Coalition, assessed that Universal Credit would be “an IT development of moderate scale, which DWP and its suppliers are confident of handling within budget and timescale”.

IBM, HP and Accenture were lined up and ready to go – with multimillion-pound contracts in hand. Government CIO Joe Harley – who doubled as DWP CIO – reassured ministers that the IT would not be a problem and that other recently successful projects were bigger than this one.

“Even in retrospect, it is difficult to explain the over-optimism,” writes Freud.

The launch was set for October 2013. For the first time, benefit claimants would be sent online to apply for the new Universal Credit. The Cabinet Office, under minister Francis Maude, had established “digital by default” as the new strategy for government. Agile development methods would reduce the risk by developing and testing the IT in “bite-sized chunks”.

“Do [IBM, HP and Accenture] know how to do it in an agile way?” Freud enquired.

“They’re up for it,” replied Malcolm Whitehouse, the IT director initially responsible for the project.

What happened next was abject failure to deliver by those suppliers, combined with poor organisation and project management by DWP, followed by years of delays and hundreds of millions of pounds written off on software that did not work or would later have to be thrown away.

Digital by default

But then came GDS, riding to the rescue on its digital steed and only too happy to pick up where the IT dinosaurs they so despised had failed, thereby hoping to prove categorically that “digital by default” was the future. Even more so, considering that DWP had been identified as the biggest IT laggard by Cabinet Office reformers.

“A team of 35 [from GDS] have produced a proof of concept that says this is the way it should be done. The previous team with 1,000 and £300m have produced something not that great,” said Maude, according to Freud’s account.

But as GDS got deeper into the weeds of implementing Universal Credit, its timetable got pushed further and further back as the true complexity of the project became clear. While happy for DWP to take the flak for its initial mistakes, GDS wanted more realistic timescales.

However, “the GDS solution doesn’t deliver what we need politically,” Freud told colleagues. The Cabinet Office had made “extravagant promises” he writes, that it could not fulfil.

And therein lies much of the controversy that surrounded the early years of GDS – a clash between the harsh realities of politics, the stubbornness and inertia of the civil service, and the reforming zeal of Francis Maude and his team. GDS eventually walked away from the Universal Credit project after disagreement over whether to keep the existing, failing system or start again from scratch following the GDS way. DWP chose the former.

The department’s permanent secretary Robert Devereux called GDS “youngsters who don’t have a clue”. Freud was told the Cabinet Office “don’t trust Robert and that’s a serious problem”. There was little love lost on either side.

Freud recounts Devereux’s critical role in Universal Credit – it’s fair to say he was less popular outside DWP, seen as the embodiment of old-school civil service ways. He had a famously frosty relationship with Labour MP Margaret Hodge, in her role as chair of the Public Accounts Committee. One exchange between the two in 2013 could have come straight from a Yes, Minister script:

“What was the Pathfinder project meant to do?” asked Hodge.

“To find a path,” replied Devereux.

Hostile informants

In a recent article for Computer Weekly, Freud said: “Computer Weekly kept a close tally of our travails, fed – we suspected – by hostile informants in the Cabinet Office. It was not surprising that Iain [Duncan Smith] came to dread the next revelation from this source.”

In his book, Freud also recalls another damaging story in Computer Weekly, attributed to “a malicious leak from the Cabinet Office”.

However, my recollection of that time was that contacts in the Cabinet Office were rarely willing to publicly badmouth DWP or the team involved with Universal Credit. GDS executives were generally very professional and declined to discuss the project beyond carefully worded statements from the press office.

The most frequent leaks came instead from inside DWP – and never for hostile or malicious reasons (and none of those sources are still working there – we went to great lengths to protect their anonymity at the time). They were people who cared deeply about the work they were doing, and often were passionate about making Universal Credit a success. But they were disturbed by the chaos in front of them, frustrated at senior managers who were covering their own backs, and annoyed by the yawning gap between their experiences and the DWP’s increasingly unrealistic public statements about progress. They felt the only way to effect change was to expose what was really going on behind the scenes.

Eventually, roll-out of a fully digital system for Universal Credit was completed in December 2018, five years later than first planned and two years after Freud stood down from frontline politics, convinced he had done as much as he could and comfortable to leave the final details to others.

That system passed its greatest test in 2020, when it processed 1.8 million claims in just six weeks from people who lost their jobs as a result of the pandemic. For all its many troubles in those early years, Universal Credit did, finally, become something of an exemplar for digital government, albeit seven years later than planned. It even runs in the cloud.

A model for government

Freud suggests that Universal Credit today has become a model for major government policies in its interplay between digital, delivery and politics. “Purely in terms of 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,” he writes.

DWP has completely restructured the way it approaches digital and IT, and has brought outsourced staff back in-house – a recognition that when your shop-front is digital, you cannot ask a third party to decide what it looks like and how it works.

In his book, Freud explains the political and parliamentary processes involved in managing such a controversial and high-profile policy in an easily understandable and engaging way. His appraisal of the mistakes made and lessons learned are honest and open, in a tale well told, as you would expect of a former Financial Times journalist.

“The fundamental mistake was over-ambition,” he writes, “based on ignorance about the implications of an interactive system which the department was simply not equipped to build.”

He adds, of government more generally: “Our governance structures are inadequate for implementing major change involving the central functions of the state” – an observation that would be considered radical if it came instead from the pen of Dominic Cummings.

When Freud stepped down from government at the end of 2016, colleagues pointed out that after a decade of driving welfare reform, he was the only consistent figure to have been there from start to finish. What does that say about government ability to manage long-term policy change?

“Bad luck and turmoil meant there were no fewer than six senior responsible owners and six programme directors in the first five years of building Universal Credit,” he writes.

“By the end of my tenure, I was seeing the third or even fourth generation of officials in the various policy areas.” Going forward, “continuity of leadership and the team as whole [should] be a central requirement.”

I also remember, back when Computer Weekly was writing so often about Universal Credit, asking the DWP press office on numerous occasions if we could interview the minister running the project, only to be told that Lord Freud was not available. He recounts in his book an encounter with two Daily Telegraph journalists that he came to regret, so perhaps that’s why – or more likely, that press officers simply didn’t want politicians to talk to specialist journalists, which is not uncommon even today.

The honesty and openness of Clashing agendas makes it a valuable insight into a policy and project that affects the lives of millions. Much was learned from Universal Credit. One can only wonder how much sooner those lessons might have been learned – and those of other major projects since – if such refreshing transparency took considerably less than a decade to achieve.

Data Center
Data Management