The Evaluation Task Force has held a conference about making government learn from evidence, which is encouraging news for a public sector that has sometimes treated evidence as something to invite, thank and leave outside with the coats.
The event, Learn Fast, Deliver Better: Innovating and Using Evidence to Reform Public Services, took place at King’s College London on 19 June 2026. More than 370 government analysts, policy professionals, academics and evaluators attended. The official report, published on 26 June, said the day focused on building the habits, tools and culture needed to use evaluation to improve public services.
This is not a small ambition. It asks government to find out whether things work, notice when they do not, and alter course without first commissioning a commemorative booklet about the original direction. The conference therefore had the atmosphere of a driving lesson held inside a vehicle that has spent several years insisting the engine noise is part of the strategy.
The Task Force’s message was that evaluation should do more than record what happened after the money has been spent and the ribbon has been cut. It should help people test ideas, improve delivery and learn in real time. This is a radical proposal for systems that have traditionally regarded learning as the moment when somebody writes a report explaining why the previous report was perfectly reasonable.
The conference examined evaluation projects involving homelessness, reoffending and victims’ support. These are serious areas in which evidence matters, not decorative subjects for a meeting room wall. The comic difficulty is that evidence is often welcomed most warmly when it arrives with good news, a laminated name badge and no request to change anything expensive.
A fictional evaluation form was clearly delighted by the new culture. It announced that at last it might be used before the decision rather than brought in afterwards to confirm that the decision had enjoyed a fascinating journey. It then requested a second form to assess whether the first form had improved outcomes.
There is the splendid institutional contradiction. Government wants to become better at learning, but must first organise a conference to announce that learning is now a priority. Still, if the paperwork has started examining its own behaviour, the revolution may be under way. The next evaluation will establish whether it has stopped filling in forms about filling in forms, while the form confirming that result waits patiently for permission to begin.
Source: GOV.UK.
