Why Government Technology Projects Fail (And What Auditors Actually Say)
Five UK digital programmes lost at least 29 years and £3 billion to supplier arrangements. The National Audit Office's diagnosis was not incompetence. It was contracts built to buy hours instead of outcomes.
Ask why government technology projects fail and you will usually get a folklore answer: bureaucracy, legacy culture, risk aversion. These are not wrong exactly, but they are not what the people who actually investigate failures say. The auditors have looked closely, repeatedly, and their findings are more specific and more actionable than the folklore.
What the UK auditors found
The National Audit Office examined five major UK government digital change programmes — the Emergency Services Network, Electronic Monitoring, Universal Credit, the National Law Enforcement Data Service, and Digital Services at the Border. Its finding: commercial approaches to working with suppliers contributed to delays totalling at least 29 years and more than £3 billion in cost increases, at least 26% above original forecast. (NAO, January 2025)
The mechanism, in the NAO's own words, is that procurement frameworks are "geared to buying inputs rather than outcomes" — contracts that specify hours and roles rather than what the system needs to actually do. A contract built this way pays for effort regardless of whether the effort produces the intended result.
The same report found a second, less-discussed cause: nobody was left to manage what had been bought. At least £14 billion is spent annually on public sector digital procurement, and only 15 people were managing relationships with government's largest digital suppliers. Commercial directors told the NAO that "very limited resource or priority is given to managing suppliers post-contract award," and that "very little information on supplier performance is available" to inform the next award.
Put those two findings together: contracts that reward hours, and almost nobody watching whether the hours produced anything. That is not a culture problem. It is a design problem, and design problems have design fixes.
What the US auditors found
The pattern is not British. The US Government Accountability Office has kept federal IT acquisition on its High-Risk List since 2015. It reports that the federal government invests more than $100 billion a year in IT, and that these investments "too frequently fail or incur cost overruns and schedule slippages while contributing little to mission-related outcomes." Of 1,881 recommendations GAO has made in this area since 2010, 463 remained unimplemented as of January 2025. (GAO-25-107852)
GAO's separate review of the government's oldest technology found the downstream cost of letting this compound: 69 legacy systems reviewed, 11 identified as most critical, ranging from 23 to 60 years old, with a combined $754 million in annual operations and maintenance cost. Eight of the eleven ran on outdated programming languages; four ran on hardware no longer supported by the vendor; seven had known cybersecurity vulnerabilities. (GAO-25-107795)
Failure to fix the contracting problem does not just cost money at the point of purchase. It compounds for decades, in systems nobody can safely touch and nobody has budgeted to replace.
The competition problem underneath both
A quieter finding sits beneath both auditors' work: buyers increasingly have fewer real choices. The European Court of Auditors found the EU-wide single-bidding rate rose from 23.5% in 2011 to 41.8% in 2021, with the average number of bidders per contract almost halving, from 5.7 to 3.2. (ECA Special Report 28/2023)
Fewer bidders means less price pressure and more dependence on suppliers a buyer cannot easily replace — which makes the outcome-versus-input contracting problem worse, not better, because a captive buyer has less leverage to insist on outcome-based terms in the first place.
What the evidence says actually helps
Not folklore. What the auditors themselves recommend, and what the clearest procurement document we have found anywhere — the US OMB's M-25-22 — requires in practice:
Specify the outcome, not the hours. This is the NAO's central finding, stated as plainly as auditors state anything.
Reserve evaluation rights and hold the data yourself. OMB guidance requires that evaluation data "should not be accessible to the vendor" and that contracts "must not prohibit agencies from internally disclosing how the vendor conducts testing or the results of testing." (OMB M-25-22)
Fund the people who manage the contract after signature. The UK's own State of Digital Government Review found only 28% of respondents believed their organisation had sufficient internal capability to monitor, track and drive supplier performance. (DSIT, January 2025) Fifteen people managing £14 billion is not a resourcing gap. It is an absence.
Price the exit before you sign, not after. Every one of the failure patterns above compounds because the buyer had already lost the leverage to demand better terms by the time the problem became visible.
The uncomfortable summary
Government technology projects do not fail because public servants are bad at technology. They fail because contracts are frequently written to measure the wrong thing, and because almost nobody is funded to check whether the right thing happened anyway. Both are fixable, in the next contract, without waiting for a culture change that may never arrive.
If you are drafting or reviewing a public sector technology contract and want these specific failure modes checked for before it is signed, that is a conversation we are glad to have.
Kaizen Spark Tech designs and delivers software, AI, automation and digital infrastructure for businesses and institutions. Every statistic here is linked to its original published source.