Learn the difference, see examples across sectors, and choose indicators that show delivery and meaningful change.
An output is a product or service delivered by an activity. An outcome is a change that follows, such as a change in knowledge, behavior, access or conditions. A course delivered is an output. Participants applying a skill at work is an outcome. Both matter, but they answer different questions.
Outputs help a team understand implementation: what was delivered, how much and to whom. Outcomes help it understand whether the intended change is occurring. Measuring an outcome does not, by itself, prove that one activity caused it.
| Level | Question | Training example |
|---|---|---|
| Input | What resources are available? | Trainer time, funding and practice equipment. |
| Activity | What does the team do? | Deliver instruction and coached practice. |
| Output | What was delivered? | Sessions delivered and participants completing the course. |
| Outcome | What changes for people or organizations? | Participants demonstrate and apply a relevant skill. |
| Impact | What wider or longer-term effects matter? | Changes in livelihoods, productivity or working conditions. |
Terminology differs across disciplines and funders. OECD’s evaluation glossary distinguishes intervention products and services from their effects, including higher-level effects. Agree definitions before aggregating reports from different teams. See the OECD evaluation glossary.
| Setting | Output | Outcome | Evidence question |
|---|---|---|---|
| Training and workforce | People complete an assessment. | People can perform the required task in practice. | Was performance assessed consistently, and could learners apply it? |
| Customer experience | Support cases receive a response. | Customers can resolve the issue and continue using the service. | Was resolution confirmed, or did the case merely close? |
| Employee development | Managers complete coaching sessions. | Employees observe changes in specific management behaviors. | Do follow-up accounts support the claimed behavior change? |
| Community services | Households receive an agreed service. | Households can access the support they need. | Who remains underserved, and why? |
| Partner or supply-chain work | Suppliers submit quarterly evidence. | The relevant practice or compliance condition improves. | Was the underlying condition verified, rather than just the form completed? |
A high output can be valuable without being sufficient. A rapid support response is useful, but the customer may still have an unresolved issue. Conversely, a low output may explain why an intended outcome has not appeared. Read delivery and change together.
An indicator turns a concept into something observable. The distinction depends on what the indicator represents, not whether it is a count, percentage or quotation. An outcome can be quantitative; an output can be described qualitatively.
| Indicator | Type in this example | Definition needed |
|---|---|---|
| Number of workshops delivered | Output | What counts as a completed workshop and which period is covered? |
| Percentage demonstrating the skill | Outcome | Which task, rubric, assessor and denominator define competence? |
| Percentage using the skill after training | Outcome | When is follow-up, how is use established, and who responded? |
| Participants reporting a barrier | Context for interpreting outcomes | What barrier categories were used and can someone report several? |
For each indicator, keep the population, unit, time period, numerator, denominator, source and owner together. A rate without a denominator is difficult to interpret; a count without a period is difficult to compare.
Start with a change statement that is specific enough to observe. “Better leadership” is too broad for a useful measure. “Managers give timely, actionable feedback that employees can use” points toward observable behavior and relevant feedback.
Individual matching is useful when the question concerns the same person over time. Repeated cross-sectional data can still describe a population trend, but changes in who responds may affect the comparison. State which design you used.
Suppose 100 people enroll in a course and 80 complete it. At follow-up, 60 completers respond and 42 say they have used the skill at work. These are different denominators and should not be merged.
| Finding | Calculation | What it tells you |
|---|---|---|
| Completion | 80 ÷ 100 = 80% | Delivery completion among enrolled participants. |
| Follow-up coverage among completers | 60 ÷ 80 = 75% | How much of the completer group is represented. |
| Reported application among respondents | 42 ÷ 60 = 70% | Application among people who answered, not necessarily all completers. |
You know that 42 people reported application. You do not know the application status of the 20 completers who did not respond. Before making a wider claim, investigate whether respondents and nonrespondents differ in relevant ways.
Store the indicator definition alongside the evidence. Where appropriate, use a stable person, account, location or partner identifier to connect forms, observations, interviews and documents across cycles. Keep access appropriate to the sensitivity of the information.
This helps a team ask why an outcome moved without rebuilding the record. AI can help classify comments and find supporting passages, but reviewers must still verify the interpretation. A software-generated summary is not a substitute for an evidence-based outcome claim.
For the wider reasoning behind a results chain, continue to the theory of change guide.
Use these existing Academy guides for the practical next step. They are suggested companion readings; follow each guide’s course navigation for the full sequence.
Explore the Loop methodology for the ongoing cycle of collection, analysis and improvement.
Its role depends on the intervention and level of analysis. A policy adopted might be an outcome of advocacy and an input to a later implementation program. State the scope.
No. Outcomes can occur in the short or medium term. The relevant timing depends on the change being studied.
No. Qualitative evidence can explain an outcome, including how people experience a change. Use a systematic collection and analysis method.
No. Outputs establish what was delivered and help explain outcomes. They are insufficient when the claim concerns a change beyond delivery.
No. You need to consider measurement quality, who was observed and other explanations for the change.
