Activity
What did people actually do? Meetings, research, repair work, training, publication, distribution, listening, testing.
SERVICE & PARTICIPATION
Why We Measure What Was Built
Attention describes a moment; evaluation asks what was delivered, what changed, what can be repeated, what failed, and what the evidence still cannot establish.
QUICK ANSWER
In one sentence
Attention describes a moment; evaluation asks what was delivered, what changed, what can be repeated, what failed, and what the evidence still cannot establish.
THE CONTROLLING CLAIM
ATTENTION IS NOT OUTPUT. APPLAUSE IS NOT CAPACITY.
TRUTH BOUNDARY
Program evaluation can distinguish activities, outputs, outcomes, effects, costs, and unintended consequences, but no single metric captures every human outcome. The seven-part measurement model and Build Measurement Receipt below are Ministry governance tools, not externally validated universal standards.
01 — OUTPUT IS NOT OUTCOME
Count the thing you delivered. Then ask what changed.
A meeting held is an activity. A guide handed to somebody is an output. A person successfully using that guide without the original builder is evidence of something closer to capacity. A durable change in the person, service, or system is an outcome. Those categories can touch each other without becoming interchangeable.
The CDC evaluation framework explicitly separates program activities, outputs, outcomes, and different evaluation types. It also warns that outcome evaluation by itself does not establish causality. GAO likewise distinguishes routine performance measurement from deeper evaluation. The Ministry adopts that distinction because attention is cheap to count and dangerously easy to mistake for consequence.
02 — SEVEN DIFFERENT QUESTIONS
A number is only useful after you know which question it answers.
Output
What usable artifact or service left the room? A guide, a repaired process, a documented source trail, a delivered service, a trained person.
Outcome
What changed for the intended person or system? Be precise about time, context, and alternative explanations.
Capacity
What can now be repeated without the original builder standing beside it? A system that dies with its founder has output but weak transfer.
Quality
Did the thing work well enough to use? Was it understandable, accessible, timely, and fit for the intended purpose?
Harm
What burden, exclusion, confusion, dependency, privacy risk, or unintended effect appeared because we built it?
Learning
What did the evidence force us to change in the next build?
03 — THE METRIC CAN EAT THE MISSION
When the score becomes the purpose, the purpose starts bending.
Page views can tell you that a page was requested. They cannot tell you whether the reader understood it, whether the source trail survived scrutiny, whether a service became easier to use, or whether a volunteer can reproduce the work without permission from the original author. Follower count can describe audience size. It cannot certify trust, truth, transfer, dignity, or usefulness.
The OECD's evaluation criteria are deliberately contextual rather than a command to collapse everything into one master number. That matters. Relevance, effectiveness, efficiency, impact, coherence, and sustainability are different questions. A metric earns its place by serving the question. The question does not exist to serve the metric.
04 — BUILD MEASUREMENT RECEIPT
Leave enough evidence for the next person to inspect the build.
- Need
- What concrete problem justified the work?
- Activity
- What did people actually do?
- Output
- What artifact or service was delivered?
- Intended user
- Who was the work for?
- Evidence of use
- What shows the intended person actually encountered or used it?
- Outcome observed
- What changed, and over what period?
- Capacity transferred
- What can another person repeat without the original builder?
- Failure
- What did not work?
- Harm or new burden
- What negative or unintended effect appeared?
- Cost
- Money, time, attention, maintenance, risk, and opportunity cost.
- What changes next
- What will be altered because of this evidence?
- What the measurement does not establish
- Name the causal, generalization, or quality claim the evidence cannot support.
05 — COMPETING INTERPRETATIONS
Measurement can clarify work. It can also become theater.
One interpretation says organizations need comparable indicators or they cannot know whether resources produced anything. Another warns that aggressive quantification can erase context and push people toward whatever is easiest to count. Both pressures are real. The Ministry's answer is not to abandon measurement. It is to keep measurement plural, contextual, challengeable, and subordinate to the actual need.
What remains unknown
No finite receipt captures every social effect. Long-term outcomes can emerge after a release ends, and many outcomes have multiple causes. Qualitative evidence can reveal consequences a dashboard misses; it can also be selective or hard to generalize. Measurement is evidence about the build, not a certificate of moral worth.
THE CONSEQUENCE
MEASURE THE THING THAT EXISTS AFTER THE ROOM GETS QUIET.
If all that remains is a screenshot of attention, count it honestly as attention. If another person can use, repair, teach, and reproduce what was built, record that too. Do not let applause impersonate capacity.
SOURCE TRAIL
- MFTAC-SRC-0059 CDC — Program Evaluation Framework, 2024
- MFTAC-SRC-0060 U.S. GAO — Performance Measurement and Evaluation: Definitions and Relationships
- MFTAC-SRC-0061 OECD — Better Criteria for Better Evaluation
CORRECTION RECEIPT
Initial publication. No post-publication factual correction recorded as of 2026-08-30. Review the public Correction Ledger for later changes.
RELATED READING