Large consultancies. Civil servants. Greed. Incompetence. IR35. A lack of STEM skills. PPE. The political merry-go-round. Scope creep. They’ve all been blamed for the regular failure of large-scale government IT projects – with their perennial ability to run billions over budget and fail to deliver functioning services on time.
(Examples abound, the ongoing Emergency Services Network imbroglio is a fine one; billions over budget, plagued by interoperability issues and described by one MP as a “shocking indictment” of the Home Office’s inability to deliver large technology projects to a clear specification whilst overseeing multiple partners.)
The National Audit Office (NAO), reviewing 25 years attempts to "deliver digital business change successfully” says it has seen a “consistent pattern of underperformance”. In a new report, it lays out six key lessons that need to be learned if future governments are going to avoid the scenario of “shifting business requirements; over-optimism; supplier performance; and lack of capability at the senior and operational level.”
Government digital transformation: 6 things must change
The government’s Major Projects Portfolio currently includes 125 projects worth £448 billion, many of which have digital elements in them. Getting their management right could save billions in tax payers' money and years by catching project overrun and scope creep early. Here's what the NAO suggests.
1: Understand the business problem
"Digital leaders told us programme teams often rush to a solution because of pressure to deliver quickly, and do not spend enough time understanding the business need, the existing system or what business improvement the programme team wants to deliver..." the NAO notes. There's an institutional malaise at play here.
As the audit office notes: "Programmes need business cases early to secure funding, and digital leaders perceive there is an incentive to show a high return on investment and give a false impression of certainty." Yet defining the scope and costs of large digital programmes require detailed exploratory work. In brief: the government needs to nurture a culture of getting detailed specifications right early, to avoid scope creep.
"Departments ask suppliers to commit to contracts without a reasonable understanding of what to deliver" - NAO, July 2021
There are a handful of disparate projects designed to improve skills on this front, but they are hardly comprehensive and tightly focussed. The Infrastructure and Projects Authority (IPA) runs a ministerial training course on how to sponsor major projects effectively but has not been commissioned to provide specific training for digital projects, the NAO notes. Earlier this year, meanwhile, the Government Digital Service created a new Digital Leaders course for senior civil servants -- but a quick look by The Stack suggests it's a mere 6.5 hours of training."Digital, data and technology leaders, who do understand these issues, need to be able to influence programme decisions to give implementing teams the best chance of success" the NAO emphasises.
2: Engage commercial partners early
Emphasising again that "more haste equals less speed", the NAO notes that departments have not typically spent enough time exploring requirements with commercial partners early in the process, noting that lengthening early engagements can de-risk projects, improve collaboration, and improve quality requirements.
See also: Changemakers Series with Sol Rashidi, Chief Analytics Officer, Estée Lauder
Currently departments "do not typically revisit, renegotiate and update contracts, except in the event of failure", the NAO emphasises; "a lapse in intelligent buying [that can] lead to adversarial relationships and poor outcomes." This will not be fixed overnight; and the NAO's deathly indictment ("departments ask suppliers to commit to contracts without a reasonable understanding of what to deliver") suggests the rot is deep.
3: The cloud does not magic away legacy issues
As part of the 2020 Spending Review, the Treasury prioritised investment in legacy IT and agreed funding for approximately £600 million to invest in modernising legacy systems, the NAO notes.
But with "failure to understand the complexity and dependencies associated with replacing legacy IT" widespread, the "government has at times found it hard to manage these transitions and complexities."
Departments typically do not have a good understanding of their IT estate and its interdependencies, and legacy systems are often poorly understood because of their age, the report notes, and some magical migration to the cloud is no panacea, even if ministers may think so.
See also: Changemakers Series - Daljit Rehal, CDIO, HMRC, on Raspberry Pis, unstructured data, and leadership
"Some applications are too old to transfer to new infrastructure and it is not always clear where the risks lie and who will remediate these applications for transfer. This continues to build up ‘technical debt’. Implementation of cloud services is not a ‘once and done’ endeavour and simply moving legacy systems into the cloud without other improvements will not resolve all the complexity, costs and risks associated with legacy systems..."
4: Be clearer about which skills to keep in-house
Setting up for effective delivery is absolutely key, with departments urged to have a clear target operating model, enterprise architecture, data model and a roadmap in place, with the former (perhaps obviously) detailing the planned "sequence of changes; the transitional states between changes; the output expected at key points in the programme; and should act as a communication tool to ensure all involved are well informed."
See also: Changemakers Series -- Tom Riordan, Leeds Council CEO, on change management & the myth of the heroic chief exec.
This is easier said and done given the large skills gaps in the civil service. Tech skills are in high demand even in the private sector -- which can pay a lot more -- and, as a result, the government more broadly "needs to set out clearly what skills it wants to develop and retain, and what skills are more efficient to contract out" the NAO emphasises.
Again, this issue runs right to the top: "Technology expertise is often under-represented at the permanent secretary and director general level compared with other types of expertise, such as policy and finance. This makes it difficult for departmental leadership to comprehend the complexities of legacy systems, data and dependencies."
5: Agile is not a panacea
Agile methods have become the default across government, but this can cause issues, the NAO suggests.
"Agile cannot solve the early stage issues that we have highlighted and can exacerbate problems when the complexity of the programme is not sufficiently understood." It highlights the rollout of the Universal Credit programme as a case in point, suggesting that the Department for Work and Pensions "focused on the IT components without fully knowing the policy and business needs".
The auditor adds that government digital transformation programmes would not be hurt by taking a long hard look at whether Agile is the right approach: "[It can be] harder than waterfall methods to manage, still need detailed planning, and can sometimes make it harder to see what progress is really being made."
6: See technology as part of a service
"The experience of industry after industry has demonstrated that just installing computers without altering the work and workforce does not allow the system and its people to reach this potential; in fact, technology can sometimes get in the way. Getting it right requires a new approach, one that may appear paradoxical yet is ultimately obvious: digitising effectively is not simply about the technology, it is mostly about the people" a 2016 report by the Department of Health and Social Care concluded succinctly.
Five years on, the NAO emphasises the same point, noting that "the government also needs to see technology as part of a service that involves people, processes and systems" and adding that there is often an "unmeasured ‘people cost’ to not modernising operational services. In the legacy landscape, large numbers of civil servants need to knit together data and find workaround solutions to compensate for the inadequacies of the legacy systems. Our Challenges in using data across government report highlighted that departments do not measure the extra people and process costs from managing inconsistent and poor-quality data, but informal estimates are that this can take 60% to 80% of some civil servants’ time."
