Outcome evaluation asks whether the changes a program intended actually happened. Learn what it can and cannot show, the steps to run one, and how to write outcome measures.
Key Takeaway
Outcomes are changes in people, not services delivered
Outcome evaluation shows whether intended changes happened, not that the program caused them, so state that limit and plan a baseline and follow-up.
What It Measures
Outcome evaluation checks whether the changes a program intended, in people's knowledge, behavior or conditions, actually happened. It does not show that the program caused them. That takes an impact design.
What It Measures
An outcome is a change in participants, not something the program delivers. Sessions held and people served are outputs. Skills gained and jobs kept are outcomes. For the full ladder, see output vs outcome vs impact.
The CDC 2024 framework describes outcome evaluation as measuring whether intended outcomes were achieved and says it cannot attribute causality. State that limit in your plan so readers do not over-read the results.
| Question | Outcome evaluation can answer | It cannot answer alone |
|---|---|---|
| Did participants change? | Yes, with a baseline and follow-up | |
| Was the change large enough to matter? | Yes, against a target | |
| Did the program cause the change? | No: needs a comparison, see impact evaluation | |
| Was the program delivered well? | No: see process evaluation |
Steps
Five Steps to Run an Outcome Evaluation
Name the outcomes. Take them from your logic model or theory of change. Keep to the two or three the program most expects to move.
Write each as a measure. Who changes, what changes, by how much and by when. See the examples below and outcome indicators.
Set a baseline. Measure before the program starts, or at enrollment. Without it, you only know where people ended up. See baseline.
Plan the follow-up. Choose the timing and how you will reach people. Losing participants is the usual weak point, so plan contact methods early.
Report with limits. Give the change, the response rate, and what the design cannot show.
Examples
Examples of Outcome Measures
These are illustrative and use invented programs.
| Program | Outcome measure |
|---|---|
| Math tutoring, a school district | Share of participants scoring at or above grade level on the end-of-year assessment, compared with their start-of-year score |
| Job training, a city workforce nonprofit | Share of completers employed 6 months after the program |
| Diabetes education, a county health department | Share of participants who report checking blood sugar as advised, at enrollment and at 3 months |
| Parent support, a family services agency | Average change on a parenting confidence scale between the first and last visit |
Avoid mixing in clinical outcome scales or patient-reported measures unless your program is clinical. Those follow their own measurement standards.
Common Mistakes
Common Mistakes
- Counting outputs as outcomes. Attendance is not change.
- Claiming cause. Say "participants improved" unless you have a comparison group.
- Too many measures. Each one needs data collection. See how to choose performance measures.
- Ignoring who dropped out. Report the response rate.
For the wider process, see Program Evaluation: Types, Steps and Examples. To draft measures with AI, use AI for indicator development or the measurement skill, then check the draft against your logic model.
Frequently Asked Questions
What is outcome evaluation?
Outcome evaluation measures whether the changes a program intended, such as knowledge, behavior or conditions, actually occurred among participants. The CDC 2024 framework notes that it cannot attribute causality; that is the job of impact evaluation.
What is the difference between outcome and impact evaluation?
Outcome evaluation shows whether intended changes happened. Impact evaluation compares results with and without the program to test whether the program caused them.
What is an example of an outcome measure?
The share of job-training participants employed six months after finishing, or the share of tutored students reading at grade level at the end of the year. Each names who, what changed and when.