As government departments allocate nearly half their technology budgets to maintaining outdated legacy systems, technical debt has evolved from an IT inconvenience into a crisis of service delivery. With an estimated 28% of central government systems now classified as legacy technology—and rising —the cost of deferring modernisation has never been clearer.
"How did we get here?" It's a question that haunts CIOs and CTOs across the public sector today. More pointedly: "How did we end up with an accidental architecture?"
Over the past two decades, there's been an almost messianic focus on building fast, failing fast, and scaling quickly. Software development at pace has trumped everything else, with the prevailing attitude being that it is better to deliver something imperfect than to pursue perfection. But as we stand on the brink of another technical revolution with generative AI, a pressing question emerges: will this supercharge our problems, or finally give us the tools to solve them?
The answer, I believe, lies in how we tackle technical debt, the silent burden that's strangling innovation across government.
What we're really talking about
Technical debt refers to the implied cost of future rework resulting from choosing an easy solution today instead of a more effective approach that would take longer. It shows up in design shortcuts, hastily written code, inadequate testing, and missing documentation.
Sometimes these tactical choices make sense: meeting an urgent public need can justify a strategic trade-off. But these tactical decisions pile up quickly, making systems more difficult to maintain, scale, or secure. The real question now: how do we know that AI-generated code, written at breathtaking speed, isn't creating even more debt for tomorrow?
The journey to agile development has created a particular challenge for public sector organisations. Digital-native companies can embrace rapid development wholeheartedly, unencumbered by legacy systems. But government departments face a more complex balancing act between delivering quick wins on citizen-facing services while carefully modernising the critical back-end systems that underpin everything.
Research by MIT's Center for Information Systems from March 2019 [Four Pathways to ‘Future Ready’ that pay off] highlights strategies for becoming future ready by transforming on two dimensions – customer experience and operational efficiency. Some focus entirely on the customer interface: building slick apps and websites while ignoring the creaking infrastructure beneath. Others disappear into what researchers called the "digitisation desert", spending months modernising core systems with nothing visible to show for it.
The right answer lies somewhere between these extremes, combining recent developments in technology, smart portfolio planning with interdependent release strategies that deliver value while building a sustainable foundation.
Scale of crisis across government
The numbers tell a sobering story. In 2021, the UK government revealed that it allocated approximately £2.3 billion annually to maintain systems burdened by technical debt – nearly 50% of its total yearly technology budget. Think about that for a moment: billions spent on keeping the lights on rather than improving services for citizens.
The situation is deteriorating. An estimated 28% of central government systems currently meet the definition of legacy technology in 2024, an increase from 26% in 2023, according to the State of Digital Government Review, published in January. In some areas, legacy technology can reach as high as 70% of systems.
These aren't abstract numbers. Across government departments, outdated systems create constant disruptions – forcing staff to revert to manual workarounds, delaying critical services, and consuming time that should be spent serving citizens.
This is what happens when we ignore technical debt: it stops being a technology problem and becomes a matter of service delivery, operational efficiency, and citizen trust.
Meanwhile, the financial drain extends far beyond maintenance costs. In 2020, the government’s Digital Economy Council estimated that total financial risk posed by legacy IT over the succeeding five years would range from £13 billion to £22 billion. Meanwhile, the UK public sector is missing out on up to £45 billion annually in potential productivity savings, according to government calculations published in January, due to outdated IT burdened by technical debt, up to 7% of total public sector spending.
The Home Office faces these challenges acutely. The department's 2024 Digital, Data and Technology Strategy explicitly acknowledge that "like other government departments, we have accrued technical debt over many years which we now need to address because it constrains our efforts to digitise the department".
Technical debt doesn't just consume budgets; it wastes human potential, which is a common challenge across large operational delivery departments like the Home Office, HMRC and DWP. When public servants spend their time fighting outdated technology instead of focusing on policy and service delivery, everyone loses.
Better managing technical debt
Managing technical debt requires sustained attention across three areas: prevention, vigilance, and planned remediation.
Prevention starts with governance. Strong architectural oversight guides design choices from the outset, providing clear standards while allowing necessary exceptions – provided they're documented, understood, and come with a plan to address any shortcuts later.
This matters more than ever with AI-generated code. Prompts need to be precise and complete, specifying clean code principles, proper documentation, and robust error handling. The temptation to accept whatever AI produces first must be resisted in favour of iterative refinement.
Ongoing vigilance means making technical debt visible in the boardroom and during portfolio planning. Different stakeholders care about different aspects – product managers naturally favour speed to market, while CFOs and chief risk officers focus on long-term costs and resilience. Use these natural tensions constructively, measuring and reporting on debt regularly to maintain focus.
Planned remediation treats maintenance as an opportunity, rather than a chore. Agile projects should allocate 25-45% of effort to refactoring. A debt-to-income ratio of less than 25% means that debt accumulates faster than you can address it; a ratio of more than 45% suggests deeper systemic issues that require urgent attention.
Building architectural maturity
The challenge isn't choosing between innovation and stability but creating the architectural maturity to enable both. This requires embedding the right behaviours, culture, and habits into development teams, treating software development as the collaborative discipline it truly is.
Current funding models make this harder than it should be. The recent State of Digital Government Review found that only one in five survey respondents felt that the current funding model enabled effective investment in and operation of digital services. Most organisations mentioned budgets that prioritise new programs at the expense of continuous improvement.
Breaking the cycle requires a different approach to investment. It’s what the policy paper A Blueprint for Modern Digital Government—published alongside the State of Digital Review, with a foreword from Rt. Hon. Peter Kyle (then Science, Innovation and Technology Secretary, now Secretary of State for Business and Trade)—calls “tailored funding models for digital products and services, legacy remediation and risk reduction, and staged, agile funding that better enables exploratory work with new technologies.”
The core question isn't around whether the government can afford to address technical debt. With 28% of red-rated legacy systems lacking remediation funding and billions at risk over the next five years, it's whether it can afford not to.
