By Tony Collins
“People burst into tears, so relieved were they that they could tell someone what was happening.”
The Institute for Government has today published one of the most incisive – and revelatory – reports ever produced on a big government IT project.
It concludes that the Universal Credit IT programme may now be in recovery after a disastrous start, but recovery does not mean recovered. Much could yet floor the programme, which is due to be complete in 2022.
The Institute’s main report is written by Nick Timmins, a former Financial Times journalist, who has written many articles on failed publicly-funded IT-based projects.
His invaluable report, “Universal Credit – from disaster to recovery?” – includes interviews with David Pitchford, a key figure in the Universal Credit programme, and Howard Shiplee who led the Universal Credit project.
Timmins also spoke to insiders, including DWP directors, who are not named, and the former secretary of state at the Department for Work and Pensions Iain Duncan Smith and the DWP’s welfare reform minster Lord Freud.
Separately the Institute has published a shorter report “Learning the lessons from Universal Credit” which picks out from Timmins’ findings five “critical” lessons for future government projects. This report, too, is clear and jargon-free.
Much of the information on the Universal Credit IT programme in the Timmins report is new. It gives insights, for instance, into the positions of Universal Credit’s major suppliers HP, IBM, Accenture.
It also unearths what can be seen, in retrospect, to be a series of self-destructive decisions and manoeuvres by the Department for Work and Pensions.
But the main lessons in the report – such as an institutional and political inability to face up to or hear bad news – are not new, which raises the question of whether any of the lessons will be heeded by future government leaders – ministers and civil servants – given that Whitehall departments have been making the same mistakes, or similar ones, for decades?
DWP culture of suppressing any bad news continues
Indeed, even as the reports lament a lack of honesty over discussing or even mentioning problems – a “culture of denial” – Lord Freud, the minister in charge of welfare reform, is endorsing FOI refusals to publish the latest risk registers, project assessment reviews and other Universal Credit reports kept by the Department for Work and Pensions.
More than once Timmins expresses his surprise at the lack of information about the programme that is in the public domain. In the “acknowledgements” section at the back of his report Timmins says,
“Drafts of this study were read at various stages by many of the interviewees, and there remained disputes not just about interpretation but also, from some of them, about facts.
“Some of that might be resolvable by access to the huge welter of documents around Universal Credit that are not in the public domain. But that, by definition, is not possible at this stage.”
Churn of project leaders continues
Timmins and the Institute warn about the “churn” of project leaders, and the need for stable top jobs.
But even as the Institute’s reports were being finalised HMRC was losing its much respected chief digital officer Mark Dearnley, who has been in charge of what is arguably the department’s riskiest-ever IT-related programme, to transfer of legacy systems to multiple suppliers as part of the dismantling of the £8bn “Aspire” outourcing venture with Capgemini.
Single biggest cause of Universal Credit’s bad start?
Insiders told Timmins that the fraught start of Universal Credit might have been avoided if Terry Moran had been left as a “star” senior responsible owner of the programme. But Moran was given two jobs and ended up having a breakdown.
In January 2011, as the design and build on Universal Credit started, Terry Moran was given the job of senior responsible owner of the project but a few months later the DWP’s permanent secretary Robert Devereux took the “odd” decision to make Moran chief operating officer for the entire department as well. One director within the DWP told Timmins:
“Terry was a star. A real ‘can do’ civil servant. But he couldn’t say no to the twin posts. And the job was overwhelming.”
The director claimed that Iain Duncan Smith told Moran – a point denied by IDS – that if Universal Credit were to fail that would be a personal humiliation and one he was not prepared to contemplate. “That was very different from the usual ministerial joke that ‘failure is not an option’. The underlying message was that ‘I don’t want bad news’, almost in words of one syllable. And this was in a department whose default mode is not to bring bad news to the top. ‘We will handle ministers’ is the way the department operates…”
According to an insider, “Terry Moran being given the two jobs was against Iain’s instructions. Iain repeatedly asked Robert [Devereux] not to do this and Robert repeatedly gave him assurances that this would be okay” – an account IDS confirms. In September 2012, Moran was to have a breakdown that led to early retirement in March 2013. He recorded later for the mental health charity Time to Talk that “eventually, I took on more and more until the weight of my responsibilities and my ability to discharge them just grew too much for me”.
Timmins was told, “You cannot have someone running the biggest operational part of government [paying out £160bn of benefits a year] and devising Universal Credit. That was simply unsustainable,”
Timmins says in his report, “There remains a view among some former and current DWP civil servants that had that not happened (Moran being given two jobs), the programme would not have hit the trouble it did. ‘Had he been left solely with responsibility for UC [Universal Credit], I and others believe he could have delivered it, notwithstanding the huge challenges of the task,’ one says.”
Reviews of Universal IT “failed”
Timmins makes the point that reviews of Universal Credit by the Major Projects Authority failed to convey in clear enough language that the Universal Credit programme was in deep trouble.
“The [Major Projects Authority] report highlighted a lack of sufficient substantive action on the points raised in the March study. It raised ‘high’ levels of concern about much of the programme – ‘high’ being a lower level of concern than ‘critical’. But according to those who have seen the report, it did not yet say in words of one syllable that the programme was in deep trouble.”
Iain Duncan Smith told Timmins that the the Major Projects review process “failed me” by not warning early enough of fundamental problems. It was the ‘red team’ report that did that, he says, and its contents made grim reading when it landed at the end of July in 2012.
Train crash on the way
The MPA [Cabinet Office’s Major Projects Authority] reviewed the programme in March 2011. “MPA reports are not in the public domain. But it is clear that the first of these flagged up a string of issues that needed to be tackled …
” In June a member of the team developing the new government’s pan-government website – gov.uk – was invited up to Warrington [base for the Universal Credit IT team] to give a presentation on how it was using an agile approach to do that.
“At the end of the presentation, according to one insider, a small number from the audience stayed behind, eyeing each other warily, but all wanting to talk. Most of them were freelancers working for the suppliers. ‘Their message,’ the insider says, ‘was that this was a train crash on the way’ – a message that was duly reported back to the Cabinet Office, but not, apparently, to the DWP and IDS.”
Scared to tell the truth
On another occasion when the Major Projects Authority visited the IT team at Warrington for the purposes of its review, the review team members decided that “to get to the truth they had to make people not scared to tell the truth”. So the MPA “did a lot of one-on-one interviews, assuring people that what they said would not be attributable. And under nearly every stone was chaos.
“People burst into tears, so relieved were they that they could tell someone what was happening.
” There was one young lad from one of the suppliers who said: ‘Just don’t put this thing [Universal Credit] online. I am a public servant at heart. It is a complete security disaster.’
IBM, Accenture and HP
“Among those starting to be worried were the major suppliers – Accenture, HP and IBM. They started writing formal letters to the department.
‘Our message,’ according to one supplier, ‘was: ‘Look, this isn’t working. We’ll go on taking your money. But it isn’t going to work’.’ Stephen Brien [then expert adviser to IDS] says of those letters: ‘I don’t think Iain saw them at that time, and I certainly didn’t see them at the time.”
At one point “serious consideration was given to suing the suppliers but they had written their warning letters and it rapidly became clear that that was not an option”.
Howard Shiplee, former head of the project, told Timmins that he had asked himself ‘how it could be that a very large group of clever people drawn from the DWP IT department with deep experience of the development and operation of their own massive IT systems and leading industry IT suppliers had combined to get the entire process so very wrong? Equally, ‘how could another group of clever people [the GDS team] pass such damning judgement on this earlier work and at the stroke of a pen seek to write off millions of pounds of taxpayers’ money?’
Shiplee commissioned a review from PwC on the work carried out to date and discovered that the major suppliers “were genuinely concerned to have their work done properly, support DWP and recover their reputations”.
In addition, when funding had been blocked at the end of 2012, the suppliers “had not simply downed tools but had carried on development work for almost three months” as they ran down the large teams that had been working on it.
“As a result, they had completed the development for single claimants that was being used in the pathfinder and made considerable progress on claims for couples and families. And their work, the PwC evaluation said, was of good quality.”
When alarm bells finally started ringing around Whitehall that Universal Credit was in trouble, IDS found himself under siege. Stephen Brien says IDS was having to battle with the Treasury to keep the funding going for the project. He had to demonstrate that the programme was on time and on budget.
‘The department wanted to support him in that, and didn’t tell him all the things that were going wrong. I found out about some of them, but I didn’t push as hard as I should have. And looking back, the MPA [Major Projects Authority] meetings and the MPA reports were all handled with a siege mentality. We all felt we had to stand shoulder to shoulder defending where we were and not really using them to ask: ‘Are we where we should be?’
‘As a result we were not helping ourselves, and we certainly were not helping others, including the MPA. But we did get to the stage between the end of 2011 and the spring of 2012 where we said: ‘Okay, let’s get a red team in with the time and space to do our own challenge.’”
The DWP’s “caste” system
A new IT team was created in Victoria Street, London – away from Warrington but outside the DWP’s Caxton Street headquarters. It started to take a genuinely agile approach to the new system. One of those involved told Timmins:
“It had all been hampered by this caste system in the department where there is a policy elite, then the operational people, and then the technical people below that.
“And you would say to the operational people: ‘Why have you not been screaming that this will never work?’ And they’d say: ‘Well, we’re being handed this piece of sh** and we are just going to have to make it work with workarounds, to deal with the fact that we don’t want people to starve. So we will have to work out our own processes, which the policy people will never see, and we will find a way to make it work.’
IBM, HP and Accenture built what’s now known as the “live” system which enabled Universal Credit to get underway, and claims to be made in jobcentres.
It uses, in part, the traditional “waterfall” approach and has cost hundreds of millions of pounds. In contrast there’s a separate in-house “digital” system that has cost less than £10m and is an “agile” project.
A key issue, Shiplee told Timmins, was that the new digital team “would not even discuss the preceding work done by the DWP and its IT suppliers”. The digital team had, he says, “a messiah-like approach that they were going to rebuild everything from scratch”.
Rather than write everything off, Shiplee wanted ideally to marry the “front-end” apps that the GDS/DWP team in Victoria Street was developing with the work already done. But “entrenched attitudes” made that impossible. The only sensible solution, he decided, was a “twin-track” approach.
“The Cabinet Office remained adamant that the DWP should simply switch to the new digital version – which it had now become clear, by late summer, would take far longer to build than they anticipated – telling the DWP that the problem was that using the original software would mean ‘creating a temporary service, and temporary will become permanent’.
“All of which led to the next big decision, which, to date, has been one of the defining ones. In November 2013, a mighty and fraught meeting of ministers and officials was convened. Pretty much everyone was there. The DWP ministers, Francis Maude (Cabinet Office minister), Oliver Letwin who was Cameron’s policy overlord, Sir Jeremy Heywood, the Cabinet Secretary, Sir Bob Kerslake, the head of the home civil service, plus a clutch of DWP officials including Robert Devereux and Howard Shiplee as the senior responsible owner along with Danny Alexander and Treasury representatives.
“The decision was whether to give up on the original build, or run a twin-track approach: in other words, to extend the use of the original build that was by now being used in just over a dozen offices – what became dubbed the ‘live’ service – before the new, and hopefully much more effective, digital approach was finished and on stream.
“It was a tough and far from pleasant meeting that is etched in the memories of those who were there…
“One of those present who favoured the twin-track approach says: There were voices for writing the whole of the original off. But that would have been too much for Robert Devereux [the DWP’s Permanent Secretary] and IDS.
” So the twin-track approach was settled on – writing a lot of the original IT down rather than simply writing it off. That, in fact, has had some advantages even if technically it was probably the wrong decision…
“It has, however, seen parts of the culture change that Universal Credit involves being rolled out into DWP offices as more have adopted Universal Credit, even if the IT still requires big workarounds.
“More and more offices, for example, have been using the new claimant commitment, which is itself an important part of Universal Credit. So it has been possible to train thousands of staff in that, and get more and more claimants used to it, while also providing feedback for the new build.”
Francis Maude was among those who objected to the twin-track approach, according to leaked minutes of the project oversight board at around this time.
Lord Freud told Timmins,
‘Francis was adamant that we should not go with the live system [that is, the original build]. He wanted to kill it. But we, the DWP, did not believe that the digital system would be ready on anything like the timescales they were talking about then …But I knew that if you killed the live system, you killed Universal Credit…”
In the end the twin-track approach was agreed by a majority. But the development of the ‘agile’ digital service was immediately hampered by a spat over how quickly staff from the GDS were to be withdrawn from the project.
Fury over National Audit Office report
In 2013 the National Audit Office published a report Universal Credit – early progress – that, for the first time, brought details of the problems on the Universal Credit programme into the public domain. Timmins’ report says that IDS and Lord Freud were furious.
“IDS and, to an only slightly lesser extent, Lord Freud were furious about the NAO report; and thus highly defensive.”
IDS tried to present the findings of the National Audit Office as purely historical.
In November 2014, the NAO reported again on Universal Credit. It once more disclosed something that ministers had not announced – that the timetable had again been put back two years (which raises further questions about why Lord Freud continues to refuse FOI requests that would put into the public domain – and inform MPs – about project problems, risks and delays without waiting for an NAO report to be published)..
Danny Alexander “cut through” bureaucracy
During one period, the Treasury approval of cash became particularly acute. Lord Freud told Timmins:
“We faced double approvals. We had approval about any contract variation from the Cabinet Office and then approvals for the money separately from the Treasury.
“The Government Digital Service got impatient because they wanted to make sure that the department had the ability to build internally rather than going out to Accenture and IBM, who (sic) they hate.
“The approvals were ricocheting between the Cabinet Office and the Treasury and when we were trying to do rapid iteration. That was producing huge delays, which were undermining everything. So in the end Danny Alexander [Lib-dem MP who was chief secretary to the Treasury] said: ‘I will clear this on my own authority.’ And that was crucial. Danny cut through all of that.”
So-called optimism bias – over-optimism – is “such a common cause of failure in both public and private projects that it seems quite remarkable that it needs restating. But it does – endlessly”.
Timmins says the original Universal Credit white paper – written long before the start of the programme – stated that it would involve “an IT development of moderate scale, which the Department for Work and Pensions and its suppliers are confident of handling within budget and timescale”.
David Pitchford told Timmins,
“One of the greatest adages I have been taught and have learnt over the years in terms of major projects is that hope is not a management tool. Hoping it is all going to come out all right doesn’t cut it with something of this magnitude.
“The importance of having a genuine diagnostic machine that creates recommendations that are mandatory just can’t be overstated. It just changes the whole outcome completely. As opposed to obfuscation and optimism bias being the basis of the reporting framework. It goes to a genuine understanding and knowledge of what is going on and what is going wrong.”
Sir Bob Kerslake, who also identified the ‘good news culture’ of the DWP as being a problem, told Timmins,
“All organisations should have that ability to be very tough about what is and isn’t working. The people at the top have rose-tinted specs. They always do. It goes with the territory.
And unless you are prepared to embrace people saying that ‘really, this is in a bad place’… I can think of points where I have done big projects where it was incredibly important that we delivered the unwelcome news of where we were on that project. But it saved me, and saved my career.”
Timmins makes good arguments for his claim that the Universal Credit programme may be in recovery – but not recovered – and that improvements have been made in governance to allow for decisions to be properly questioned.
But there is no evidence the DWP’s “good news” culture has changed. For instance the DWP says that more than 300,000 people are claiming Universal Credit but the figure has not been audited and it’s unclear whether claimants who have come off the benefit and returned to it – perhaps several times – are being double counted.
Timmins points out the many uncertainties that cloud the future of the Universal Credit programme – how well the IT will work, whether policy changes will hit the programme, whether enough staff will remain in jobcentres, and whether the DWP will have good relations with local authorities that are key to the delivery of Universal Credit but are under their own stresses and strains with resourcing.
There are also concerns about what changes the Scots and Northern Irish may want under their devolved powers, and the risk that any ‘economic shock’ post the referendum pushes up the volume of claimants with which the DWP has to deal.
Could Universal Credit fail for non-IT reasons?
“In seeking to drive people to higher earnings and more independence from the benefits system, there will be more intrusion into and control over the lives of people who are in work than under the current benefits system. And there are those who believe that such an approach – sanctioning people who are already working – will prove to be political dynamite.”
The dire consequences of IT-related failure
It is also worth noting that Universal Credit raises the stakes for the DWP in terms of its payment performance, says Timmins.
“If a tax credit or a Jobseeker’s Allowance payment or any of the others in the group of six go awry, claimants are rarely left penniless in the sense that other payments – for example, Housing Benefit in the case of Jobseeker’s Allowance or tax credits, – continue.
“If a Universal Credit payment fails, then all the support from the state, other than Child Benefit or disability benefits not included within Universal Credit, disappears.”
This happened recently in Scotland when an IT failure left hundreds of families penniless. The DWP’s public response was to describe the failure in Scotland as “small-scale”.
What a report.
It is easy to see how much work has gone into it. Timmins has coupled his own knowledge of IT-related failure with a thorough investigation into what has gone wrong and what lessons can be learned.
That said it may make no difference. The Institute in its “lessons” report uses phrases such as “government needs to make sure…”. But governments change and new administrations have an abundance – usually a superfluity – of confidence and ambition. They regard learning lessons from the past as putting on brakes or “nay saying”. You have to get with the programme, or quit.
Lessons are always the same
There will always be top-level changes within the DWP. Austerity will always be a factor. The culture of denial of bad news, over-optimism about what can be achieved by when and how easily it can be achieved, over-expectations of internal capability, over-expectation of what suppliers can deliver, embarking on a huge project without clearly or fully understanding what it will involve, not listening diligently to potential users and ridiculously short timescales are all well-known lessons.
So why do new governments keep repeating them?
When Universal Credit’s successor is started in say 2032, the same mistakes will probably be repeated and the Institute for Government, or its successor, will write another similar report on the lessons to be learned.
When Campaign4Change commented in 2013 that Universal Credit would probably not be delivered before 2020 at the earliest, it was an isolated voice. At the time, the DWP press office – and its ministers – were saying the project was on budget and “on time”.
The National Audit Office has highlighted similar lessons to those in the Timmins report, for example in NAO reports on the NPfIT – the NHS IT programme that was the world’s largest non-military IT scheme until it was dismantled in 2011. It was one of the world’s biggest IT disasters – and none of its lessons was learned on the Universal Credit programme.
The NPfIT had an anti-bad news culture. It did not talk enough to end users. It had ludicrous deadlines and ambitions. The politicians in charge kept changing, as did some of programme leaders. There was little if any effective internal or external challenge. By the time it was dismantled the NPfIT had lost billions.
What the Institute for Government could ask now is, with the emasculation of the Government Digital Service and the absence of a powerful Francis Maude figure, what will stop government departments including the DWP making exactly the mistakes the IfG identifies on big future IT-enabled programmes?
In future somebody needs the power to say that unless there is adequate internal and external challenge this programme must STOP – even if this means contradicting a secretary of state or a permanent secretary who have too much personal and emotional equity in the project to allow it to stop. That “somebody” used to be Francis Maude. Now he has no effective replacement.
It’s also worth noting in the Timmins report that everyone seems to be a victim, including the ministers. But who are perpetrators? Timmins tries to identify them. IDS does not come out the report smelling of roses. His passion for success proved a good and bad thing.
Whether the direction was forwards or backwards IDS was the fuel that kept Universal Credit going. On the other hand his passion made it impossible for civil servants to give him bad news – though Timmins raises questions about whether officials would have imparted bad news to any secretary of state, given the DWP’s culture.
Neither does the DWP’s permanent secretary Robert Devereux emerge particularly well from the report.
How it is possible for things to go so badly wrong with there being nobody to blame? The irony is that the only people to have suffered are the genuine innocents – the middle and senior managers who have most contributed to Universal Credit apparent recovery – people like Terry Moran.
Perhaps the Timmins report should be required reading among all involved in future major projects. Competence cannot be made mandatory. An understanding of the common mistakes can.
Thank you to FOI campaigner Dave Orr for alerting me to the Institute’s Universal Credit reports.
Thanks also to IT projects professional John Slater – @AmateurFOI – who has kept me informed of his FOI requests for Universal Credit IT reports that the DWP habitually refuse.
Update 18.00 6 September 2016
In a tweet today John Slater ( @AmateurFOI ) makes the important point that he asked the DWP and MPA whether either had held a “lessons learned” exercise in the light of the “reset” of the Universal Credit IT programme. The answer was no.
This perhaps reinforces the impression that the DWP is irredeemably complacent, which is not a good position from which to lead major IT projects in future.