What the DAC Criteria Are
The OECD Development Assistance Committee published the original five evaluation criteria in 1991: relevance, effectiveness, efficiency, impact, and sustainability. In 2019, they revised the definitions, added coherence as the sixth criterion, and updated the language to reflect modern evaluation practice. These criteria are the most widely used framework for structuring evaluations in international development.
They are not a checklist. They are a menu. Every evaluation should use some of them. No evaluation should try to use all of them equally. Choosing which criteria to prioritize, and writing sharp evaluation questions for each, is the real skill.
At a Glance
| Criterion | Core Question | Added/Revised |
|---|---|---|
| Relevance | Is the program doing the right thing? | Original (1991), revised 2019 |
| Coherence | Does the program fit with other interventions? | New in 2019 |
| Effectiveness | Is the program achieving its objectives? | Original (1991), revised 2019 |
| Efficiency | Are resources being used well? | Original (1991), revised 2019 |
| Impact | What broader difference does the program make? | Original (1991), revised 2019 |
| Sustainability | Will the benefits last? | Original (1991), revised 2019 |
The Six DAC Evaluation Criteria
Relevance
Definition: Is the intervention doing the right things? Does it respond to the needs, policies, and priorities of beneficiaries, the country, and the global context? Is it still relevant as circumstances change?
Example evaluation questions:
- To what extent does the program address the priority needs identified by target communities, and how have those needs changed since the program was designed?
- How well does the program design align with the national development strategy and sector policies?
Common pitfall: Treating relevance as a formality. Many evaluations answer "yes, the program is relevant" without actually examining whether needs have shifted since design. The 2019 revision explicitly emphasizes responsiveness to changing contexts. Ask whether the program is relevant now, not just whether it was relevant when it started.
Coherence
Definition: How well does the intervention fit? Is it compatible with other interventions in the same country, sector, or organization? This includes internal coherence (alignment between the intervention and the organization's own policies) and external coherence (complementarity with what other actors are doing).
Example evaluation questions:
- To what extent does the program complement or duplicate the activities of other organizations working on nutrition in the same districts?
- How well does the program align with the organization's global strategy and other country-level programming?
Common pitfall: Confusing coherence with coordination. Coherence is about strategic fit, not whether organizations hold joint meetings. Two programs can coordinate well and still be incoherent (pursuing contradictory approaches to the same problem). Two programs can have zero coordination but be highly coherent (different entry points to the same goal). Also note that coherence is the newest criterion and many evaluators are still figuring out how to operationalize it. Keep the questions focused and do not let this criterion expand into a full mapping of every actor in the sector.
Effectiveness
Definition: Is the intervention achieving, or is it likely to achieve, its objectives and results? This includes differential effects across groups.
Example evaluation questions:
- To what extent has the program achieved its intended outcome of increasing household dietary diversity among target beneficiaries?
- What factors have enabled or hindered the achievement of program results, and how do outcomes differ between urban and rural areas?
Common pitfall: Measuring outputs instead of outcomes. Effectiveness is not "did we deliver the training?" (that is an output). Effectiveness is "did the training change practices?" (that is an outcome). The 2019 revision also emphasizes looking at differential effects: who benefited and who did not. If your evaluation reports average outcomes without disaggregation, you are missing a key part of effectiveness.
Efficiency
Definition: How well are resources being used? Is the intervention delivering results in an economic and timely way? This includes comparing the costs and benefits of the intervention to alternative approaches.
Example evaluation questions:
- How does the cost per beneficiary of this program compare to similar programs using alternative delivery models?
- Could the same results have been achieved with fewer resources, or could more results have been achieved with the same resources?
Common pitfall: Reducing efficiency to a single cost-per-beneficiary number without context. A program that costs $50 per beneficiary is not automatically more efficient than one that costs $200 per beneficiary. The $200 program might deliver more durable or deeper outcomes. True efficiency analysis requires comparing costs to results, not just comparing costs. See cost-effectiveness analysis for methods.
Impact
Definition: What difference does the intervention make? This covers higher-level effects, both intended and unintended, positive and negative. The 2019 revision broadened impact beyond the narrow "attributable effect" definition to include the wider change the intervention contributes to.
Example evaluation questions:
- What broader changes in food security at the district level can be observed since the program started, and what is the program's plausible contribution to these changes?
- What unintended effects (positive or negative) has the program had on non-target populations or on gender dynamics within target communities?
Common pitfall: Conflating impact with effectiveness. Effectiveness is about your program's intended outcomes. Impact is about the bigger picture. A training program might be effective (participants learned new skills) but have limited impact (those skills did not translate to income changes at the community level). Impact evaluations typically require more rigorous designs, larger samples, and longer timeframes. Most program evaluations should focus on effectiveness and only address impact if the budget and design support it. See How to Choose Evaluation Methodology for guidance on matching your design to your questions.
Sustainability
Definition: Will the net benefits of the intervention continue after funding ends? This includes financial, institutional, social, and environmental sustainability.
Example evaluation questions:
- To what extent have local institutions developed the capacity and resources to continue program activities without external support?
- What is the likelihood that the behavioral changes observed among beneficiaries will persist two years after the program ends?
Common pitfall: Asking about sustainability too early. A mid-term evaluation of a 5-year program cannot meaningfully assess whether benefits will last after the program ends. It can assess whether the program is building toward sustainability (capacity building, handover planning, local ownership). Frame sustainability questions appropriately for the evaluation timing. Also avoid the vague question "Is the program sustainable?" Instead, specify what you expect to be sustained: services, behaviors, institutions, policies.
Choosing Which Criteria to Use
Do not use all six. Here is how to choose.
Start with the evaluation purpose. A formative mid-term evaluation and a summative final evaluation serve different purposes and should emphasize different criteria.
| Evaluation type | Recommended criteria | Why |
|---|---|---|
| Mid-term formative | Relevance, Effectiveness, Efficiency | Focus on course correction. Too early for impact or sustainability. |
| Final summative | Effectiveness, Impact, Sustainability | Focus on results and lasting change. Relevance is less actionable at the end. |
| Thematic/sector review | Coherence, Relevance, Effectiveness | Understand how programs fit together across a sector. |
| Real-time evaluation | Relevance, Effectiveness | Fast feedback during implementation. Efficiency, impact, sustainability come later. |
Limit yourself to 3-4 criteria. With 1-2 evaluation questions per criterion and a realistic budget, 3-4 criteria produce focused, actionable evaluations. Five or six criteria spread resources thin and produce superficial answers to all questions rather than deep answers to the important ones.
Total evaluation questions: aim for 6-10. Each criterion should have 1-3 questions. More than 10 questions total means your evaluation is trying to do too much. Prioritize ruthlessly. Every question you add increases data collection costs, analysis time, and report length. Write your questions into an evaluation matrix to verify that each one has a clear data source, method, and analysis plan.
ALNAP Criteria for Humanitarian Evaluations
Humanitarian evaluations often add two additional criteria from ALNAP (Active Learning Network for Accountability and Performance):
Coverage: Did the intervention reach the intended population, especially the most vulnerable?
Connectedness (or Coordination): Was the intervention linked to longer-term recovery and development efforts?
These supplement the DAC criteria for humanitarian contexts where reaching affected populations and linking relief to recovery are critical concerns. Use them when evaluating emergency response, refugee programming, or disaster risk reduction.
Common Mistakes
Mistake 1: Using all six criteria in every evaluation. This is the most common error. A terms of reference that lists all six criteria with three questions each produces 18+ evaluation questions, which no evaluation can answer well within a typical budget. Choose 3-4 criteria. Justify the choice. Leave the rest out.
Mistake 2: Writing vague evaluation questions. "Is the program relevant?" is not an evaluation question. "To what extent does the program address the priority needs identified by community health committees in the target districts?" is. Vague questions produce vague answers. Every question should specify what you are evaluating, for whom, and against what standard.
Mistake 3: Treating the criteria as independent boxes. The criteria interact. A program can be highly effective but not efficient (achieving results at excessive cost). A program can be relevant but not coherent (addressing real needs but duplicating what others are doing). Your analysis should note these connections. Do not write six isolated chapters in the evaluation report.
Mistake 4: Adding coherence because it is new. Coherence was added in 2019 and many terms of reference now include it reflexively. Only include coherence if it matters for your evaluation purpose. If the evaluation is about one program's internal results, coherence may not be a priority. If the evaluation is about how multiple programs work together in a sector, coherence is essential.
Mistake 5: Confusing the DAC criteria with an evaluation framework. The criteria tell you what to evaluate. They do not tell you how. You still need to choose methods, designs, and data collection approaches for each question. The criteria structure the evaluation TOR. The methodology section of the TOR specifies how to answer the questions. Use the Evaluation Designer to connect criteria to appropriate methods.
Mistake 6: Ignoring unintended effects under the impact criterion. The 2019 revision explicitly includes unintended effects, both positive and negative. A nutrition program that improves dietary diversity (intended) but increases women's workload (unintended negative) has produced both. Your impact assessment should look for both. Ask stakeholders what changed that was not expected. Do not limit data collection to your predefined indicators.