What are the OECD-DAC criteria?
The Development Assistance Committee (DAC) of the Organisation for Economic Co-operation and Development (OECD) published a set of criteria for evaluating development and humanitarian work. Many donors and NGOs use them to structure evaluations. They give evaluators and project teams a shared language.
The criteria are not a formula. They are six questions that help you look at a project from six sides. This article gives each one in plain words. The exact official wording belongs to the OECD, so use their current text when you write terms of reference.
What do the six criteria ask?
| Criterion | Plain question | Sample evaluation question |
|---|---|---|
| Relevance | Is the project doing the right things? | Do the activities match what the community said it needed? |
| Coherence | Does it fit with other work? | Does it complement or duplicate what other agencies do in the area? |
| Effectiveness | Is it reaching its goals? | To what extent did the planned outcomes happen, and for whom? |
| Efficiency | Are resources used well? | Were activities delivered on time and at a reasonable cost? |
| Impact | What difference does it make? | What lasting changes, planned or not, can we see? |
| Sustainability | Will the benefits last? | Will communities keep the benefit when funding ends? |
How do you read each criterion in practice?
Relevance. Look back at the needs assessment. Did the project respond to the needs of the people it meant to serve, and did those needs change during the project? A well-run activity that nobody needs is not relevant.
Coherence. This looks outward. Check how the project fits with government services, other NGOs and your own other projects. It can reveal gaps and overlaps.
Effectiveness. Compare results with your targets. Ask also who benefited and who did not. An average can hide the fact that women, older people or people with disabilities were left out.
Efficiency. This is about how well you turned money, time and staff into results. It does not mean the cheapest option. A cheap activity that does not work is not efficient.
Impact. Ask about bigger and longer-term change. This is the hardest to measure, because other factors also cause change. Be modest about what you can claim.
Sustainability. Ask who will keep things going, and with what money, skills and rules. Look for local ownership and for risks that could undo the gains.
How do you choose which criteria to use?
You rarely need all six. Choose by purpose and timing:
- Write down why the evaluation is happening. For example, to decide whether to scale up, to account to a donor or to learn what to change.
- Match the purpose to the criteria. A midterm review often focuses on relevance and effectiveness, since you can still change course. A final evaluation may add impact and sustainability.
- Limit the questions. Three to five questions per criterion is a lot. Fewer, sharper questions give better answers.
- Agree the evidence. For each question, note what data you will use: your indicators, interviews, group discussions or records.
What mistakes should you avoid?
- Copying generic questions. Write questions about your project.
- Scoring without evidence. If you give a rating, show what it rests on.
- Treating impact as proof. One project rarely proves a change across a population.
- Forgetting accountability. Include the views of affected people, not only staff and partners.
- Starting too late. Good evaluations need baseline data, so plan them when you write the MEAL plan.
How do you turn a criterion into usable questions?
Take one criterion and write questions in three steps. First, state what you want to know in plain words. Second, add the group or place that matters. Third, name the evidence.
For example, under effectiveness: “Did women farmers in the northern districts change how they store grain after the training, and what explains the difference between districts?” Evidence could be a before-and-after survey, interviews with farmers and trainer records.
Share the draft questions with the field team and with community members where possible. They often spot questions that do not make sense locally. Rank the questions by importance, and drop the ones that nobody will act on. An evaluation with eight sharp questions will usually give more value than one with thirty vague ones. Check your donor’s terms of reference as well, since some donors fix the criteria or the questions in advance.
Where EMPOBase fits
EMPOBase does not run evaluations or score projects against the OECD-DAC criteria. It helps with the evidence an evaluator will ask for: results frameworks tied to Programs, indicator actuals in Monitoring, Risk & Learning records and an After Action Review for project closure. Library indicators are reference text, so teams still review and adapt each one. Totals are counts of records, not unique people, which an evaluator should note when reading reach figures. See the indicator library for examples.