Four questions that separate an impact report from an impact document.
Search for guidance on impact reporting and you will find a great deal of help with the report. Templates. Structures. What to include. How to design it, how to tell the story, which framework to align to, how to make the numbers land.
Almost none of it addresses the question that determines whether the exercise was worth the money: did anything happen differently because of it?
This is not a gap in the guidance. It is a gap in the thing the guidance describes. Most impact reporting is genuinely not designed to change a decision — it is designed to account for one already taken. Both are legitimate. The problem is that organisations fund the first while expecting the second, and then conclude their measurement is inadequate when the real issue is that their reporting was never wired to anything.
Four questions will tell you which one you have built.
1. Name the decision
Before the next reporting cycle, write down the specific decision this report is meant to inform. Not "improve our programmes" — a decision, with a verb. Whether to continue the second cohort. Whether to move delivery in-house. Whether to renew the partner contract. Whether to change the eligibility criteria.
If no decision can be named, the report is an account, not an input. That is a fine thing to be, but it should be resourced and scheduled as an account and it should stop being described as learning.
Most teams find this question harder than expected. That difficulty is the finding.
2. Name the owner
Who takes that decision? A named role, not a committee-shaped noun. "The leadership team" is usually a way of saying nobody.
Here the mandate–responsibility mismatch tends to surface. The person accountable for the outcome frequently has no authority over the thing that drives it — they can be held responsible for retention while having no say over selection, or for sustainability while procurement sits elsewhere. Reporting into that gap produces frustration on a schedule.
If the owner of the decision is not the audience of the report, you have a distribution problem that no amount of better analysis will solve.
3. Name the date
When is that decision taken, and does the evidence arrive before it?
This single question retires an enormous amount of impact reporting. Annual reports inform annual decisions. If the choice about the second cohort is made in March and the evidence lands in June, the evidence is a post-mortem regardless of quality. It will be read. It will be appreciated. It will change nothing.
Matching cadence to decision is usually cheaper than improving methodology, and it moves the needle far more. A rough number in February beats a rigorous one in June, every time, for a March decision.
4. Name what they can do about it
Suppose the evidence says the programme is underperforming. What is that body empowered to do? Reallocate budget? Halt delivery? Change the design? Or note it and ask for a further update?
If the honest answer is the last one, the reporting loop has no actuator. You have built a very good thermometer and connected it to nothing. This is the layer that gets skipped most often, because it is the only one that requires taking authority from somewhere and giving it somewhere else.
What changes when you do this
Organisations that run these four questions properly tend to find the same three things.
They are measuring more than they need. A surprising share of collected indicators map to no decision at all. Retiring them frees capacity and nobody misses them.
They are measuring the wrong cadence. The most decision-relevant evidence is often needed quarterly or at gateway moments, and is currently produced annually because that is when the report is due.
And they discover that the binding constraint was never evidence. It was that no forum was obliged to look at it, and no role was accountable for responding. Which is an organisational design problem, and responds to organisational design.
The test, restated
An impact report that changes a decision has four properties: a named decision, a named owner, a date that precedes the decision, and a body with the authority to act. Miss any one and the loop is open.
Most impact reporting misses at least two, and it is almost never the measurement that is at fault.