Over the past few years, I’ve been encouraged by the growing support for a ‘What Works’ approach to tackling homelessness. Policymakers, funders, frontline organisations, and journalists often reach out to discuss how we can use evidence to drive better decision-making. It’s heartening to see this shift—there is a clear appetite for using data and research to improve outcomes, and a recognition that good intentions alone are not enough.

But as enthusiasm for evidence-informed policy grows, the danger of oversimplification also increases. The phrase ‘What Works’ is now widely used, but not always in ways that reflect its true meaning. We hear some projects described as ‘test and learn,’ but without clear hypotheses, comparison groups, or structured evaluation, in which the learning is often informal rather than systematic. We see interventions referred to as ‘evidence-based’ that draw on internal reports, rather than a robust independent evaluation. And we find assumptions that simply gathering data—any data—means being data-led, without questioning whether the right questions are being asked or the data is being interpreted correctly.

This is understandable. The desire to act quickly is strong, especially in an area as urgent as homelessness. But if we want to make real progress, we need to move beyond broad commitments and get clear on what a true evidence-based approach looks like in practice—and what it doesn’t.

The Centre for Homelessness Impact was created to embed a ‘What Works’ approach at the heart of how we tackle and prevent homelessness, and we've drawn powerful inspiration from the evidence-based medicine movement's transformation of healthcare. It wasn't always a smooth ride in medicine, and there are still challenges, but it's now the standard – and that's the kind of change we want to see in the homelessness  sector.

What ‘What Works’ really means

At its core, a ‘What Works’ approach is about using the best available evidence to inform decisions while being open to improving that evidence over time. It’s not about waiting for perfect data before acting, nor is it about rigidly following research without considering context. Instead, it’s about shifting from assumptions to knowledge, from good ideas to tested solutions. And it’s about continuing to experiment and evaluate.

This is about more than routinely using data and evidence in your work. Data can be useful, misleading, or entirely meaningless depending on how it’s collected and used. The same is true about evidence. So an evidence-based approach to ‘what works’ in homelessness involves designing and implementing interventions or policies grounded in rigorous, up-to-date research and data. By rigorous evidence we mean data that has been systematically gathered and analysed to answer a specific question. 

Drawing from evidence-based medicine, a true ‘What Works’ approach requires three fundamental commitments:

1. Moving from ‘test and learn’ to rigorous testing

The phrase ‘test and learn’ has become common in policymaking, and with good reason—it signals a willingness to adapt and improve. The challenge is ensuring that testing is structured enough to provide meaningful insights.

For example, if an organisation introduces a new intervention and sees a positive outcome, can we be sure the intervention caused that change? Could it have happened anyway? Would another approach have worked better?

A robust test requires:

A clear hypothesis—What specific outcome do we expect the intervention to produce?

A comparison group—What would have happened without the intervention, allowing us to isolate its effect?

Predefined outcome measures—What specific indicators will be used to determine the intervention's success or failure, established before the intervention begins?

Independence of evaluation design, management, and delivery—Is the design, management, and delivery of the evaluation conducted objectively, free from influence by those delivering the intervention or with a stake in a particular outcome?

Declaration and management of potential conflicts of interest—Are all potential conflicts of interest identified, disclosed, and effectively managed to minimise bias throughout the process?

The most rigorous way to answer these questions is through randomised controlled trials (RCTs), but other strong methods—such as quasi-experimental designs—can also provide valuable insights. What matters is that the intervention has a clear theory of change and evaluation is built in from the start, rather than added as an afterthought.

2. Asking the right questions

Bad research often starts with the wrong question. In medicine, thousands of studies examine whether a particular treatment works—but far fewer explore whether the problem could have been prevented in the first place.

In homelessness very few rigorous evaluations of common interventions look at prevention - just cast an eye at our Intervention tool that provides an overview on the strength of the evidence on 19 of the most common homelessness interventions. 

Much of the existing research asks: Who uses X or Y service and what are their needs? 

But often, the more transformative question is:

Is the intervention desirable and cost effective? How does it compare with other interventions trying to do the same thing?

For instance, consider temporary accommodation (TA). Many studies explore the costs of temporary accommodation, and the financial burden it is placing on local authorities. But we could gain more impact by focusing more on experiments that can help reduce the flow of people into TA or bypassing it altogether. Or ensuring children in TA are not adversely affected in other ways that evidence suggests will have life-long effects such as poor educational and health outcomes. 

A true ‘What Works’ approach prioritises the questions that lead to meaningful change—not just the ones that measure the prevalence of a social issue or challenge.

‍3. Not all studies are equal: understanding research quality

Before citing a statistic, ask: Where does this data come from, and how reliable is it? It’s useful to group research in broad categories:

Gold standard: Systematic reviews and large-scale studies using robust methods (e.g., randomised controlled trials or longitudinal research).

Useful but limited: Case studies, small sample surveys, and expert opinion—valuable but should not be mistaken for universal truths.

Red flag: Advocacy-driven reports that don’t disclose their methodology. If you can’t see how the data was collected, be cautious.

Be wary of reports that make sweeping claims about ‘what works”’ without clear evidence of causality. Look for research that controls for other factors or, even better, uses experiments to isolate cause and effect.

Whether you’re a policy-maker practitioner or someone who’s reporting or funding homelessness interventions, when faced with new research you should ask yourself: Who conducted the research and how was it funded? Are claims about causation backed by strong evidence? Are key limitations or uncertainties acknowledged; is there a broader context to help interpret findings?

When it comes to reporting on homelessness, the right data can illuminate solutions, challenge misconceptions, and shape policy. But misinterpreted or cherry-picked statistics can do the opposite—reinforcing myths, misleading the public, or even just missing the real story. Beware of big numbers without context - large numbers grab headlines, but they can mislead if not framed correctly.

A true ‘What Works’ approach looks at the full body of evidence, using systematic reviews and meta-analyses to understand the bigger picture.

What ‘What Works’ doesn’t mean

Just as important as defining what a ‘What Works’ approach’ is is clarifying what it isn’t.

1. It’s not about ignoring experience

Some worry that focusing on data means sidelining the expertise of frontline workers or people with lived experience. In reality, the best approaches integrate three types of knowledge:

Scientific evidence—What do rigorous studies tell us?

Practical expertise—What do professionals on the ground observe?

Lived experience—What do people who have experienced homelessness say about what helped?

A robust ‘What Works’ approach values all three.

2. It’s not about waiting for perfect evidence before acting

Some fear that an evidence-based approach leads to paralysis—that we must wait for gold-standard studies before making decisions. That’s not how it works. In medicine, doctors must often act with incomplete information, using the best available evidence while continuing to learn. The same applies to homelessness policy. We don’t need to wait for perfect studies before acting, but we do need to:

• Use the strongest evidence available
• Track outcomes rigorously
• Be ready to adapt or stop interventions if they don’t work

Conclusion: a commitment to learning, not just action

The growing enthusiasm for evidence-led decision-making is a hugely positive shift. But if we want to make real progress, we have an opportunity to go beyond slogans and ensure that ‘What Works’ isn’t just a phrase—it’s a practice. That means moving from informal ‘test and learn’ approaches to structured testing. It means asking the right questions, not just the easiest ones. And it means looking at the full body of evidence, and scaling (or not) interventions based on a systematic approach. Most of all, it means a willingness to change course when the evidence demands it. 

The road to evidence-based practice is rarely straightforward, as medicine's experience shows, but the progress in healthcare proves that by embracing evidence and being persistent, we can transform the way we work to end homelessness. This is the heart of a true ‘What Works’ approach—and it's the only way we will make lasting progress in ending homelessness.

At the Centre for Homelessness Impact, we advocate for better use of evidence in policymaking service provision and journalism alike. If you’re ever unsure about a statistic or want to explore the latest high-quality research, we’re here to help.

• Lígia Teixeira, CEO of the Centre for Homelessness Impact