The fast and the curious: A hypothesis-led approach to tackling the temporary accommodation and homelessness crisis

The challenge: a surge in demand for Temporary Accommodation.

In September 2023, Royal Borough of Greenwich, along with councils across London and the country, was facing a homelessness and Temporary Accommodation (TA) crisis.

Following the pandemic, the cost of living crisis, and a downturn in affordable housing, the number of households living in TA in London had risen to 66,000. With it the cost of TA for the 32 London councils had increased by a collective £90 million over the previous 12 months alone.

The scale of this housing crisis has an awful effect on residents. The number of residents in need of housing support continues to grow, and the length of time some residents are having to spend in unsuitable accommodation is growing with it.

Back in September 2023, there were an unprecedented number of households in TA in Greenwich, a worrying increase in the use of hotels as temporary accommodation (from 5 hotel rooms per night in September 2022, to 269 a year later), and the single largest overspend the council had ever forecast.

While our housing services continued to work tirelessly and looked for new ways to try to intervene as early as possible before peoples’ situations worsened, the scale of the crisis meant we had to manage its symptoms as well as its cause. Namely, reducing the booking frequency and unit cost of:

• Hotel rooms
• Emergency overnight accommodation (self-contained temporary accommodation which we rent on a nightly basis).

Forming a targeted approach to tackle the crisis

This was the point at which our digital team became involved in designing a response. Tasked with simultaneously understanding the problem space and delivering solutions as quickly as possible, we proposed taking a hypothesis-led approach to solving these knotty problems.

A diagram showing the 4 steps of the hypothesis-led approach and the links between them This means that, rather than following the usual double-diamond approach of understanding the problem before developing and delivering solutions, we started with credible potential solutions, and identified the smallest ‘hypotheses’ we could test to prove or disprove their validity.

Benefits of taking a hypothesis-led approach

Leaping to solutions like this feels inherently risky, and there are many instances in which the risks of this method would not be worth the reward. In the context of this crisis, however, we believed it was our best approach because:

1. The human and financial impacts of the TA crisis meant we had to move fast

Starting with hypotheses and working backwards rapidly accelerated the problem-solving process, and allowed us to start delivering value immediately while learning more about the rapidly changing situation which we were yet to fully understand.

2. We had a wealth of knowledge within service staff who already understood the issues and had suitable ideas to tackle them

We convened over 40 staff with subject matter expertise from a dozen teams across three directorates, to hypothesise interventions we could implement to reduce the human and financial impacts of the crisis.

3. We were confident in our mitigations against the risks of ‘leaping to solutions’.

At the hypothesis-generation session, we captured a comprehensive list of assumptions we needed to prove in order to be confident each hypothesis would work. This lets us test assumptions before jumping to implementation, significantly reducing the chances of us overlooking potentially negative unintended consequences of implementing these solutions.

From ideas to action: How we brought our hypotheses to life

We quickly prioritised our hypotheses in smaller workshops with subject matter experts, then tested them rapidly in priority order, discarding the ones that proved invalid, and carrying forward the ones that were borne out by evidence, bringing in new ones from the backlog as capacity permitted. Those that proved successful, for example making direct allocations of social housing to eligible homeless households who have spent years in TA, have since been scaled up. Others, such as the idea of offering to buy properties from landlords whose tenants approached us with eviction notices, proved invalid. But we were able to discard them quickly and focus our energies on the most valuable ideas.

To facilitate this process, we worked hard in the first few months of the programme to improve our data, until we had wide-ranging, trustworthy, and responsive reports of key metrics. Close-to-real-time data allowed us to quickly test the impact of our hypotheses on a small scale, and a reliable impact model can extrapolate the projected impact of our hypothesis tests over the longer term. This tells us, with a reasonable level of confidence, the number of households we can help out of TA and into secure accommodation and the extent to which we can reduce TA unit cost if we scale up our hypotheses. It also allows for an evidenced-based comparison of those benefits with any tradeoffs that might come with scaling up our hypothesis. For example, how changes to allocations strategy may affect non-homeless households bidding for social housing. This evidence base helps to mitigate the inherent risks involved in being hypothesis-led by increasing our confidence in the solutions we choose to implement.

Colleagues from a range of teams and departments then set about implementing the hypotheses we were confident would succeed and deliver value quickly. For us as a digital team, this meant using a mixture of our data and design skills to deliver the following in collaboration with operational teams:

• A live pricing data tool which has helped the service reduce our average hotel room unit cost by 21%
• An improved new-build mobilisation process, which allows us to let new council housing twice as quickly as we would previously been able to.

Finally, what impact did we see and what learnings have we taken as we continue to move forward?

Our rapid, hypothesis-led response delivered some remarkable outcomes. Just over a year after the programme started:

• We have reduced the number of households in hotel B&B accommodation from a peak of 289 down to ~100.
• Families applying as homeless are being assessed up to 6 times faster than they previously had been.
• We have supported hundreds of families and dozens of single households into better quality social and or temporary accommodation, including supporting dozens of older TA households into sheltered accommodation.
• Alongside all this, the programme’s interventions have had a positive impact on the council’s financial position of £2.7 million in financial year 23/24 and a further £2.4 million in 24/25.

What we’ve learned

Reflecting on these achievements a year after our first hypothesis generation workshop, three key reflections come to mind

1. We can be confident to take risks in the approach we take when we’re building around strengths

We knew we had the strengths necessary to provide an effective response to the crisis – subject matter expertise, committed, dedicated, and skilled staff, and strong leadership across the organization. The approach we took, although it was new and sometimes uncomfortable, was designed to enhance those strengths to deliver impact as quickly as possible.

2. The nature of the hypothesis-led approach is well suited to crisis-response

As we have discussed, hypothesis-led problem solving supports the fast delivery of impact, which was a key consideration in the response to this crisis. It also, however, lends itself to a move towards a more sustainable long-term approach further down the line. Thanks to the hypothesis testing carried out so far, we have a high level of confidence in determining the most successful and impactful interventions at our disposal. As such, we’ve streamlined our programme of work down to a small number of key strategic pieces of work which we are confident can be resourced, delivered, and achieve our goals.

3. Last but far from least, collaboration is key

The causes and symptoms of this crisis are complex in their nature. It is not something housing, regeneration, welfare, finance, or digital experts could solve on their own. Together, however, they form a group with the expertise necessary to identify the solutions, and the skills needed to deliver them. Behind the strategic theory, the key to the success of our hypothesis-led approach so far has been its ability to bring experts from across the organisation together and foster genuine collaboration.

When recently announcing an innovation fund for test and learn teams initially focused on TA, Pat McFadden, Chancellor of the Duchy of Lancaster, perfectly summed up the need for innovative approaches to knotty challenges: ‘It needs an appetite for risk. Some things won’t work properly first time. But if we are terrified of failure we will never innovate and we’ll carry on doing what we’ve always done.’ As ‘test and learn’ approaches to TA become more prevalent, we’re keen to search for and share best practice with other local authorities. If you are facing similar challenges, or this article has piqued your interest, please do get in touch and we can continue to iterate, improve, and innovate together.