Evaluation research is the systematic study of whether a program, product, policy, or service achieved its goals and whether the time, money, and effort behind it were worth it. It is also called program evaluation. It is defined by its purpose, not by one method.
Most teams already run some form of it. One training manager compares test scores before and after a course. Another product team asks why users abandon setup. Meanwhile, a health agency checks whether a vaccination drive reached the people it targeted.
The difference between a casual check and real evaluation research is rigor: clear questions, a baseline, the right data, and honest reporting. This guide covers the types, methods, steps, and examples you need to run it well.
What is evaluation research?
Evaluation research is a form of applied research that judges the merit, worth, or effect of something that already exists or is being tested. It answers practical questions: Did it work? For whom? At what cost? What should change next?
Three traits separate it from other research:
- Purpose: It exists to support a decision, such as continue, fix, scale, or stop.
- Setting: It happens inside real programs and organizations, not in a lab.
- Audience: It must satisfy stakeholders, including funders, managers, staff, and participants.
Because it serves decisions, evaluation research borrows methods from social science but adds a clear “so what” at the end. A finding that does not change a decision is a report, not an evaluation.
How is evaluation research different from similar terms?
Evaluation research is often confused with basic research, monitoring, and market research. The table below separates them by the question each one asks.
| Term | Core question | Typical output |
|---|---|---|
| Evaluation research | Did this program or product achieve its goals, and was it worth the cost? | Findings and recommendations for a decision |
| Basic research | Why does something happen? | New theory or knowledge |
| Applied research | How can we solve a specific practical problem? | A solution or method (evaluation is one type) |
| Program monitoring | Is the work happening on schedule? | Activity counts and progress dashboards |
| Market research | What do customers want or think? | Demand, preference, and segment insight |
Monitoring tells you what you did. Evaluation tells you whether it mattered. Many programs have plenty of the first and almost none of the second.
When should you do evaluation research?
You should evaluate at every stage of a program, not only at the end. Early work shapes the design, mid-course work fixes delivery, and end-stage work proves results.
Plan the evaluation before the program starts. If you wait until the program ends, you will not have a baseline, which is the starting measurement that lets you show change. Without one, even a large improvement is hard to attribute to your work.
Good triggers include:
- Launch: You are piloting a new program, feature, or policy.
- Midpoint: Participation or results look different from the plan.
- Funding review: A funder, board, or executive wants proof of impact.
- Scale decision: You must choose between expanding, changing, or ending the effort.
Why does evaluation research matter?
Evaluation research matters because it replaces opinion with evidence when resources are limited. It protects good programs from being cut and exposes weak ones before they absorb more money.
- Better decisions: Leaders see what works, what does not, and where a project stands.
- Continuous improvement: Feedback from staff, clients, and participants shows what to fix next.
- Accountability: Funders, boards, and regulators get proof that intended benefits reached the target audience.
- Capacity building: Results reveal skill gaps and resource needs that planning documents miss.
- Fairness: Breaking results out by group shows whether a program helps everyone it was meant to help.
What are the types of evaluation research?
The main types are formative, process, outcome, impact, and summative evaluation. Each fits a different stage and answers a different question. Teams often combine two or three in one study.
Formative evaluation
Formative evaluation happens before or during early delivery, and its goal is to improve the design. It checks whether the idea is clear, feasible, and well matched to the audience.
- Typical question: Do participants understand the program, and what would make it work better?
- Typical method: Needs assessment, concept testing, pilot feedback.
Process evaluation
Process evaluation checks how a program is delivered. It asks whether activities followed the plan, who took part, and what got in the way. It explains why results look the way they do, which makes it the most useful diagnostic type when outcomes disappoint.
- Typical question: Was the program delivered as designed, and who actually took part?
- Typical method: Attendance records, staff logs, and short participant check-ins.
Outcome evaluation
Outcome evaluation measures changes in the people or conditions the program targets, such as knowledge, skills, behavior, or satisfaction. It compares results with the baseline and with the stated goals. It also looks for unintended effects, good and bad.
- Typical question: Did participants gain the knowledge, skill, or behavior the program promised?
- Typical method: Pre/post surveys and follow-up surveys at fixed intervals.
Impact evaluation
Impact evaluation goes one step further and asks what changed because of the program, not just after it. It needs a comparison group or a strong design to rule out other causes. It is slower and costlier, so reserve it for decisions with high stakes. A credible comparison group might be people on a waitlist, similar sites without the program, or participants matched on age and starting skill.
Summative evaluation
Summative evaluation judges overall success once a program ends or reaches a milestone. It pulls together outcomes, costs, and lessons so leaders can decide whether to renew or retire the effort.
- Typical question: Did the program deliver enough value to justify continuing?
- Typical method: Combined outcome data, cost data, and stakeholder interviews.
Educators will recognize the same split in educational evaluation and learning evaluation.
Which evaluation research methods should you use?
Use quantitative methods to measure how much changed, qualitative methods to learn why, and mixed methods when the decision needs both. Most strong evaluations mix them.
Quantitative methods
Quantitative methods produce numbers you can compare and track. They are the backbone of outcome and impact work.
- Surveys and questionnaires: Structured questions, often on rating scales, sent to participants.
- Pre-test and post-test: The same questions asked before and after the program.
- Administrative records: Attendance logs, test scores, sales data, or health records you already hold.
- Experiments: Participants are assigned to a program group or a comparison group.
Qualitative methods
Qualitative methods capture motivation, context, and experience that numbers hide. Participants explain what happened in their own words. These methods take longer to run and analyze, and the findings are harder to generalize, so use them to explain results rather than to prove them.
- Interviews: One-on-one conversations that surface personal reasons and barriers.
- Focus groups: Small group discussions that show shared views and disagreement.
- Observation: Direct notes on how a service or product is actually used.
- Case studies: Detailed accounts of a few participants, sites, or events.
Mixed methods
Mixed methods pair a survey with follow-up interviews, or open-ended comments with rating scales. A rating drop tells you something went wrong. A handful of interviews tells you what.
Evaluation designs compared
The design you pick sets how confidently you can claim the program caused the change.
| Design | How it works | Strength of evidence | Best used when |
|---|---|---|---|
| Post-only | Measure after the program | Low | Quick satisfaction checks |
| Pre/post | Measure before and after | Moderate | Most training and service programs |
| Comparison group | Compare participants with similar non-participants | Stronger | A fair comparison group exists |
| Randomized trial | Assign people to groups by chance | Strongest | High-stakes decisions, adequate resources |
How do you conduct evaluation research step by step?
You conduct evaluation research by defining the program, agreeing on questions, gathering credible evidence, and acting on what you find. The CDC’s program evaluation framework, built for public health, lays out six steps and five standards that work in any sector. The steps are to assess the context, describe the program, focus the evaluation questions and design, gather credible evidence, generate and support conclusions, and act on findings.
Here is a practical version of that sequence.
- Define the program and the decision.
Write down the goals, the target audience, and the decision the results will inform. If nobody can name the decision, pause here. - Map the logic.
Draw a one-page logic model that links resources, activities, outputs, and outcomes. This shows which results are realistic to measure. - Write focused evaluation questions.
Limit yourself to three to five questions. Each one should be answerable with data you can actually collect. - Choose indicators and a design.
For each outcome, set an indicator, a data source, a timing, and a target. Pick the pre/post or comparison design that fits your budget. - Collect baseline and follow-up data.
Use identical questions at each wave so you can compare results. Track response rates as you go. - Analyze, report, and act.
Compare results with the baseline and the targets, explain gaps with qualitative evidence, state limitations, and assign owners to the recommendations.
What are examples of evaluation research questions?
Good evaluation questions are specific to the program type and the evaluation type. The examples below show how the same logic adapts across settings. All scenarios are illustrative.
| Setting | Evaluation type | Sample question |
|---|---|---|
| Workforce training | Outcome | Did participants apply the new skill on the job within 30 days? |
| Public health outreach | Process | Which neighborhoods did the program reach, and which did it miss? |
| Software onboarding | Formative | At which setup step do new users stop, and why? |
| School tutoring | Impact | Did tutored students improve more than similar students who were not tutored? |
| Customer service change | Summative | Did satisfaction and resolution time improve enough to justify the cost? |
For more question wording, see these program satisfaction survey questions.
Worked example: A job-readiness program
Imagine a nonprofit that runs a 12-week job-readiness program. It surveys participants at enrollment, at graduation, and six months later.
- Process question: Did 80% of enrolled people attend at least 10 of the 12 sessions?
- Outcome question: Did the share of participants employed rise between enrollment and the six-month follow-up?
- Qualitative follow-up: What barriers kept non-completers from finishing?
If employment rises from 30% to 55%, the change is 25 percentage points. That number is useful, but without a comparison group, it does not prove the program caused the gain. A careful report says so.
How do you measure whether a program worked?
You measure success by comparing each indicator with its baseline and its target, then checking whether the change is large, consistent, and plausibly caused by the program. An indicator is a specific, observable measure of an outcome, such as “percentage of graduates employed at six months.”
Use these checks when you review results:
- Change from baseline: Subtract the starting value from the later value. Report the raw difference, not just a percentage of a percentage.
- Target met: Compare the result with the target you set before the program began, not one you picked afterward.
- Subgroup differences: Break results out by site, cohort, or demographic group to see who benefited.
- Response rate and bias: Report how many people answered and who may be missing, since people who finish a program often answer more favorably.
- Cost per outcome: Divide total cost by the number of people who reached the outcome when funders ask about value.
Funders are asking for outcomes over activity counts. For deeper guidance, see this framework on proving impact to funders and this guide to measuring program impact.
How do you choose the right evaluation approach?
Start with the decision, then match the type, method, and design to it. The decision sets the evidence you need, and the evidence sets the cost.
- Choose formative evaluation if the program is new or changing and you can still adjust the design.
- Choose process evaluation if results are weaker than expected and you need to find where delivery broke down.
- Choose outcome evaluation if you must show participants changed against the goals.
- Choose impact evaluation if the decision involves large budgets or policy and you can build a credible comparison.
- Choose mixed methods if the audience wants both the numbers and the story behind them.
Match effort to stakes. A low-cost pre/post survey is enough for a short workshop. A multi-year public program may justify a comparison group.
How do you keep evaluation research ethical and credible?
You keep it credible by protecting participants, staying independent, and being transparent about method and limits. The CDC framework lists ethics, rigor, transparency, independence, and relevance among its standards, and they apply well beyond public health.
In practice, tell participants why you are collecting data and how you will use it. Get consent, keep responses confidential where you promised to, and store personal data securely. When the team that runs the program also runs the evaluation, add an outside reviewer or publish the full method so readers can judge the work for themselves.
What mistakes weaken evaluation research?
The most common mistakes are starting too late, measuring activity instead of change, and hiding limitations. Each one erodes the credibility that makes evaluation worth doing.
- Skipping the baseline: With no starting point, you cannot show change.
- Changing questions between waves: Different wording makes pre and post results incomparable.
- Counting outputs as outcomes: “We trained 400 people” is an output. A rise in skills or employment is an outcome.
- Ignoring non-response: A 20% response rate from the happiest participants can mislead.
- Claiming causation too soon: Without a comparison group, say the results are associated with the program.
- Collecting data nobody uses: Each question should map to an evaluation question and a decision.
How can QuestionPro support evaluation research?
QuestionPro Market Research Software supports the data side of evaluation research: building the instrument, reaching the right people, and analyzing results.
- AI Survey Agent: It drafts a questionnaire from a plain-language brief, including question types and logic, which speeds up pre/post instruments.
- Audience panels: Pre-screened respondents help when you need an outside comparison group or a hard-to-reach segment.
- Data Quality AI: It flags bot, AI-generated, and gibberish responses before analysis.
- Smart Dashboards and Text AI: They produce cross-tabs and group open-ended comments by topic and sentiment, so you can compare subgroups and explain results.
- Communities and InsightsHub: Always-on research communities support repeat measurement, and InsightsHub stores past studies for later comparison.
You can explore the full toolset on the Market Research Software page. The platform helps you collect and analyze evidence. The evaluation design is still yours to get right.
Evaluation research turns effort into evidence
Evaluation research is how a team learns whether its work made a difference. It starts with a decision, uses a baseline, picks methods that fit the stakes, and reports results honestly, including the limits.
Programs that measure only activity can describe how busy they were. Programs that evaluate can show who changed, by how much, and why. That is the evidence that wins renewals, guides improvements, and earns trust.
Frequently Asked Questions (FAQs)
Its main purpose is to help people decide what to do next. It measures whether a program, product, or policy met its goals, who benefited, what it cost, and what should be continued, changed, scaled, or stopped.
Process evaluation looks at how a program was delivered, including attendance, fidelity to the plan, and obstacles. Outcome evaluation looks at what changed for participants afterward. Together, they explain not only whether results improved but why.
It can be either, and strong studies use both. Surveys, tests, and records measure how much changed. Interviews, focus groups, and observation explain why it changed. The right mix depends on the evaluation question and the decision.
Market research studies what customers want, think, or might buy. Evaluation research judges whether something you already built or ran worked. Both use surveys and interviews, but evaluation compares results against goals and baselines.
Yes. A small team can write three to five focused questions, collect a baseline, run an identical post-survey, and report results with limitations. A simple pre/post design is not causal proof, but it gives funders useful, honest evidence.



