The art of fixing something important: lessons from programmes that work (and those that don’t)

Designing public or social programmes that truly change lives is deceptively hard. Policies are often well-intentioned, heavily funded, and expertly planned, yet still fall short of meaningful impact. Others, modestly resourced but laser-focused, quietly deliver lasting value.

When you look closely, success isn’t random. It comes down to how clearly a programme defines its problem, how honestly it measures its effects, and how faithfully it sticks with implementation when the political spotlight moves on. As Malcolm Sparrow reminds us in The Regulatory Craft, governments often “fixate on compliance or coverage” when what’s needed is focus on real harm reduction.

Across numerous evaluations and case studies, clear patterns emerge: when design, implementation, and measurement align, results follow. When they don’t, programmes stall. Not because people don’t care, but because their systems pull in different directions.

Focus matters: when you chase everything, you fix nothing

The first failure mode is fuzzy focus. When policy teams try to fix multiple problems at once, measurement and delivery lose coherence.

Take the UK Troubled Families Programme. Early phases were criticised for shifting goals and output-heavy reporting (e.g., “families engaged”) rather than outcomes (e.g., “sustained reduction in truancy or re-offending”). Later iterations tightened definitions and data dashboards, letting local teams target specific, measurable harms. The result was steadier improvement and more credible evidence (National Audit Office, 2016).

Finland’s Basic Income Experiment offers a counterpoint. Its design was ruthlessly clear: test whether unconditional income improves employment and wellbeing. The employment results were neutral, but the wellbeing data were strong, showing that clarity of purpose doesn’t guarantee success but does ensure learning (Kangas et al., 2021).

Funding shapes behaviour: incentives decide what gets done

Every funding model sends a signal. Output-based contracts reward activity; outcome-based ones reward results, but both can misfire if misaligned.

In Australia’s Jobactive programme, audits found that short-term employment incentives drove providers to focus on easier-to-place clients, undermining equity (Australian National Audit Office, 2017). Likewise, the UK’s Work Programme paid on sustained job outcomes, but risk-averse providers still “parked” harder cases (Department for Work and Pensions, 2014). Both show how measurement and payment architecture quietly shape what happens on the ground.

True performance frameworks need a blend: payments for process quality and verified outcomes, plus clear definitions of what “success” means to participants, not just to funders.

Implementation craft: policy is 10% design, 90% delivery

Good design dies fast without committed delivery. Successful programmes invest in backbone organisations, capable coordination, and the messy work of learning.

The Harlem Children’s Zone and StriveTogether initiatives show how durable backbones, consistent data, and community ownership produce measurable gains in education and wellbeing (Kania & Kramer, 2011; StriveTogether, 2020). Similarly, New Zealand’s Whānau Ora model demonstrates the power of devolved commissioning that allows communities to define success through their own values, although evaluators note that mismatched expectations between funders and communities make outcome evidence complex to aggregate (Te Puni Kōkiri, 2019).

Compare that with short-lived pilots that lose momentum when political will or funding shifts. Finland’s basic income trial ended abruptly once the government changed, representing a design success but an implementation casualty (Kangas et al., 2021).

What you measure defines what you learn

Measurement is not neutral, it defines what counts as reality. Too often, teams measure what’s easy (spend, participants, training hours) instead of what matters (lives improved, harm reduced, confidence gained).

StriveTogether’s collective impact model works because partners commit to shared measurement across organisations. Everyone tracks the same few indicators, so progress is visible and learning continuous (StriveTogether, 2020).

In contrast, fragmented programmes with multiple reporting systems drown in data but lack insight. You can’t prove change or improve it without a baseline, shared definitions, and longitudinal follow-up.

Measurement should not be an audit exercise; it should be a learning loop that feeds data back into design and delivery to enable real-time course correction.

Divergent perspectives: when success means different things

Many programmes fail not from poor design but from mismatched expectations. Sponsors may define success as fiscal savings, designers as fidelity to evidence, and beneficiaries as lived improvement.

When these aren’t reconciled, trust fractures. In Whānau Ora, community outcomes such as cultural connection or whānau cohesion can’t always be expressed in government spreadsheets. In employment or health interventions, agencies may prize “completion rates” while participants value autonomy or dignity.

Malcolm Sparrow describes this as the difference between managing for appearances and managing for substance (Sparrow, 2000). The challenge for modern public managers is to align these definitions by making space for multiple truths while maintaining a coherent system of accountability.

Collective impact: coordination is an intervention

Collective impact isn’t just collaboration; it’s structure. Successful initiatives build five ingredients: a common agenda, shared measurement, mutually reinforcing activities, continuous communication, and a backbone organisation (Kania & Kramer, 2011).

When these are real, the model works — as seen in StriveTogether’s education networks or New Zealand’s Border Executive Board, where cross-agency alignment produced demonstrable efficiency and service improvements (Public Service Commission, 2023). When they’re superficial (no shared metrics, no governance authority), “collaboration” becomes a talking point rather than a delivery mechanism.

True collective impact takes time, trust, and resourcing for coordination, not just project funding.

Failure signals you can see coming

Across contexts, early warning signs are remarkably consistent:

• No shared understanding of the harm being addressed.
• No baseline data or inconsistent measurement frameworks.
• Funding cycles that reward outputs or short-term activity.
• Weak governance or absence of a coordinating “backbone.”
• Political churn without protected implementation funding.
• Limited community participation in design or evaluation.

When these appear, the programme’s probability of failure rises sharply. It’s not because people lack competence, but because the system lacks alignment.

Final thought

The hard truth is that most programmes do not fail because they were badly designed. They fail because design stopped at the whiteboard. Ideas were approved, money was spent, and the machinery moved on before the real work, the learning, adapting, and redesigning, could take hold.

True impact comes when design, delivery, and measurement stay connected long enough for insight to grow and trust to deepen. It means staying with the problem, not just announcing the solution. As Sparrow reminds us, the craft of public problem solving is not to fix everything. It is to fix something important, and to fix it well, over and over, until the system itself begins to learn.

References

• Australian National Audit Office. (2017). Design and monitoring of the jobactive employment services program.Canberra, ACT: ANAO.
• Department for Work and Pensions. (2014). The Work Programme: Evaluation of payment model and outcomes. London: UK Government.
• Kania, J., & Kramer, M. (2011). Collective impact. Stanford Social Innovation Review, 9(1), 36–41.
• Kangas, O., Jauhiainen, S., Simanainen, M., & Ylikännö, M. (2021). The basic income experiment 2017–2018 in Finland: Preliminary results. Helsinki: Ministry of Social Affairs and Health.
• National Audit Office. (2016). Troubled Families Programme: Progress review. London: NAO.
• Public Service Commission. (2023). Methodology and evaluating collective impact: Border Executive Board case study.Wellington, NZ: Te Kawa Mataaho Public Service Commission.
• Sparrow, M. K. (2000). The regulatory craft: Controlling risks, solving problems, and managing compliance. Washington, DC: Brookings Institution Press.
• StriveTogether. (2020). Theory of Action: Building cradle-to-career outcomes through collective impact. Cincinnati, OH: StriveTogether.
• Te Puni Kōkiri. (2019). Whānau Ora review: Tipu Matoro ki te Ao. Wellington, NZ: Ministry of Māori Development.