How can organisations establish whether an intervention caused improvement?
Published by Ceilyte™ · Last reviewed: 16/11/2026An intervention's delivery does not establish its effectiveness. To make a credible causal claim, compare outcomes against a pre-intervention baseline and a suitable counterfactual, then investigate whether other explanations could account for the observed change.
Activity, output, outcome and attribution
Running 40 training sessions is an activity. The number of participants who completed them is an output. An increase in the proportion of staff who feel safe to speak up is an outcome. Neither activity nor outcome alone establishes that the training caused the change.
Methodology: test the causal claim
- Define the expected change and how the intervention is supposed to produce it.
- Establish a baseline measured before delivery, using comparable definitions and population coverage.
- Identify a credible comparison group or comparison period. Consider what would likely have happened without the intervention.
- Check whether the sequence and timing of the change are consistent with the proposed mechanism.
- Examine alternative explanations, including leadership change, restructuring, staffing changes, case mix and concurrent initiatives.
- Triangulate independent data sources and contradictory accounts. Report the strength of the attribution claim separately from the observed improvement.
Worked example
Illustrative scenario. A trust reports that speaking-up confidence increased from 54% to 64% after training. Over the same period, a comparable department without the training improves from 55% to 63%. The difference in changes is two percentage points, not ten. That comparison is informative, but does not establish causation unless the groups are sufficiently comparable and other explanations are addressed.
Limitations: what if there is no control group?
Ceilyte does not always have access to a control group. In these circumstances, it examines the timeline, mechanism, pre-existing trends, independent documentary evidence, feedback and alternative explanations. Triangulation can make a contribution claim more credible, but cannot automatically establish a counterfactual or quantify a causal effect. Conclusions must reflect the quality and completeness of evidence.
Primary sources
- HM Treasury — The Magenta Book
- HM Treasury — Quality in policy impact evaluation
- National Audit Office — Evaluating government spending
Related resources
- How can public bodies independently assess value for money?
- How can regulators, inquiries and boards verify recommendations?
- Evidence assurance glossary
Commission an independent evidence review
Discuss your programme, recommendation or evidence requirements with Ceilyte.
Book a 30-minute call