Over the break, I read a New Year piece by Tim Harford that stayed with me. Instead of urging readers towards self-improvement, the kind of big promises January seems to demand, he reflected on something far less fashionable: maintenance. 

His argument is that many of the things that matter most fail not because they lack ambition, but because we stop looking after them. Not through a single decision, but by neglect: small choices, year after year, that slowly weaken what once worked. Harford also makes a point we often miss. Good maintenance is not mindless routine. It is skilled work. It is the ability to notice early warning signs, to keep systems usable, and to prevent avoidable failure.

As I read it, I found myself thinking less about personal resolutions and more about the state of our social security safety net.

At its best, social security functions as national infrastructure. It absorbs shocks that are common in real lives: job loss, illness, disability, caring responsibilities, rising housing costs. It reduces the risk that these events turn into long-term harm. It protects children from instability that can shape outcomes for years. When support works as intended, people recover faster, return to work sooner, and remain connected to family and community. When it fails, problems compound, and homelessness becomes more likely.

But a safety net only works if it is maintained.

Since the early 2010s, the capacity of the system to prevent problems early has steadily narrowed. This is not because prevention fell out of favour in principle. Successive governments have continued to commit to it. And yet sustained spending pressures and rising demand pushed resources towards managing immediate need. Over time, early intervention, delivery capability, and the infrastructure that supports learning and adaptation became harder to protect. Maintenance suffered as a result, not through a single decision, but through repeated trade-offs made under pressure. 

When prevention and maintenance weaken, spending does not disappear. It shifts. Homelessness shows this clearly. Temporary accommodation is the result of late intervention. It is costly, disruptive, and particularly damaging for children and future generations. Councils in England now spend around £2.8 billion a year on temporary accommodation, a 25 per cent rise in the past year and more than twice what was spent five years ago. We see similar trends in Scotland, Wales and Northern Ireland. It shows how reactive spending on crisis response absorbs resources that might otherwise support early intervention and stable housing.

In the run-up to Christmas, the UK government published a series of ambitious strategies on homelessness, child poverty, and financial insecurity. Taken together, they reflect a welcome recognition that these challenges are connected, and that upstream prevention matters.

But strategies alone do not deliver change. What will determine whether this moment leads to different outcomes is whether delivery, integration, and learning are treated as core work rather than afterthoughts.

At present, in our work at the Centre for Homelessness Impact, we see a consistent gap on these fronts. National government is asking local systems to take on greater responsibility and collaborate more deeply, often before governance arrangements are mature or shared accountability is clear. It expects data-led decision making where data is incomplete or inconsistent. But when devolution runs ahead of the infrastructure needed to support it, pressures do not disappear. They re-emerge downstream, in homelessness services, stretched local authority budgets, and poorer outcomes for families.

This is why the role of central government still matters, even in a world where local areas rightly take more ownership. The centre is uniquely placed to set direction and invest in the foundations that local systems cannot build alone, including shared data infrastructure, analytical capability, and hub-and-spoke models that connect national goals with local delivery. It also has a critical role in ensuring a more joined-up approach across central government departments, so that action taken for legitimate reasons in one public service does not inadvertently increase homelessness or other pressures elsewhere. In the absence of these capabilities, even well-designed, evidence-based policies will struggle to take root.

This is where maintenance alone is not enough. Maintaining a system that was not designed for current expectations, or that cannot make effective use of new technologies at scale, will not produce better results. Care without construction preserves limitations.

But the opposite error is just as common. New initiatives layered onto weak foundations rarely deliver lasting change. Without attention to delivery, coordination, and learning, innovation becomes churn: activity without progress.

As we enter 2026, the real task is to do both.

We need to maintain what already works: approaches that show results, relationships built over time, and the everyday capabilities that allow services to function. At the same time, we need to be clear where the infrastructure is not fit for purpose, and invest deliberately in building it.

In homelessness prevention, that means developing systems that support early action rather than defaulting to crisis response. It means shared data and learning environments so local areas can see what is happening and adjust course. It means strengthening delivery capability so evidence can be implemented, not just published. It also means investing in evaluation and cost-effectiveness, so resources flow towards what actually reduces harm.

This work will not be easy, but it is where credible ambition now sits. As the year begins, our resolution is a clear one, and one we hope others will recognise as their own: to take maintenance as seriously as we take reform, and to build the foundations that make change last.

‍

• Lígia Teixeira is Chief Executive of the Centre for Homelessness Impact

‍