The Cabinet Office has announced £4.7 million for 40 new evaluation projects designed to test whether public policies and services produce their intended results. The work spans artificial intelligence, employment, education, public health and violence prevention, using methods that range from randomised trials to natural experiments.
What is being funded?
The open-competition awards form part of the fifth phase of the Evaluation Accelerator Fund. The wider phase has a stated value of £6.5 million for 2026–27, including additional support for What Works Centres and mayoral strategic authorities. The 40 projects are the largest single group in the announcement.
Which projects stand out?
Examples include a Newcastle trial of an AI-supported approach to school absence, a natural-experiment evaluation of heat-health alerts, an evaluation of AI in employment support and work on access to GCSE examination centres. Funding amounts vary by project, reflecting different questions and methods.
Why evaluate policies experimentally?
A programme can be popular or well intentioned without causing the outcome credited to it. Random allocation, comparison groups and carefully designed quasi-experiments can separate an intervention’s effect from wider trends. Qualitative work can then explain implementation, experience and context that a headline effect size may miss.
Does the announcement prove the ideas work?
No. These are evaluation awards, not results. Some projects will test feasibility or build data infrastructure; others will estimate impact or value for money. Findings may be positive, negative or inconclusive. Publishing methods and results, including null findings, will determine how useful the portfolio becomes.
What happens next?
Many projects are expected to produce shareable findings by March 2027, although publication may follow later and some matched-funded work can run longer. Readers should look for preregistered methods where appropriate, transparent limitations and links to final reports. Evidence-based government depends on using results, not merely commissioning studies.
Primary sources: the Cabinet Office Phase 5 project summaries and the competition guidance.
Reporting basis: official information published on 21 September 2026. Illustrative image: photo by Stephen Dawson on Unsplash.