June 20, 2026 - BY Admin
If you work in monitoring and evaluation, you have almost certainly been handed a terms of reference that ask you to assess a project against the OECD-DAC criteria: relevance, coherence, effectiveness, efficiency, impact, and sustainability, and you have almost certainly never been formally taught what any of them actually mean in practice. They sit in the document like a checklist you're expected to already understand. So you do your best, you organise the report under six headings, and you hope the structure does the talking.
Here's what we've learned from sitting on both sides of these evaluations: the six criteria aren't really headings. Each one is a question a donor is asking quietly in the background, and the subtext of that question matters far more than the label printed on top of it. A report that answers the labels but misses the subtext reads as competent but shallow. A report that answers the real questions reads as the work of someone who has actually run an evaluation rather than someone formatting one.
The criteria aren't a bureaucratic hurdle. They're a shared language that lets a donor compare very different projects against a common set of questions about quality. Used well, they're a thinking tool. Used badly, they become six boxes to fill.
This is the version we wish someone had given us early on : for each criterion, the surface question, what the donor is genuinely probing, a concrete example of the difference, and a practical way to handle it in your own work.
A quick note on "six," not "five"
If you trained on this framework more than a few years ago, you learned five criteria. In 2019, the DAC revised them, adding coherence as the sixth. It's the one most teams still fumble, partly because it's newest, partly because it's the least intuitive. A report that quietly ignores coherence signals to a knowledgeable reader that the evaluator is working from an outdated mental model.
Relevance"Was this the right thing to do?"
On the surface, relevance asks whether the project responded to a real need. That's easy enough to assert, and most evaluations assert it confidently. The catch is that donors aren't only asking whether the need existed when the project was designed, they're asking whether you genuinely understood it, and whether your understanding held up as the context shifted around the project.
Did you design for the population, or did you design for the proposal? In humanitarian and development settings especially, needs move. A project that was highly relevant at approval can become quietly irrelevant eighteen months later when a drought breaks, a market collapses, displacement patterns change, or a policy shifts the landscape. An honest relevance section tracks that movement rather than freezing the analysis at design.
For example: a livelihoods project designed to boost maize yields may have been perfectly relevant when prices were strong. If prices crashed mid-implementation and the real need became market access, an evaluator who writes "the project remained highly relevant throughout" has missed the point and lost credibility for everything that follows.
How to handle it: treat relevance as a moving target. Document the need at design, then test whether it still held at each major phase. Where context shifted, say so plainly and assess how well the project adapted.
Coherence"Did it fit with everything else going on?"
This is the newest criterion, and it's really two questions wearing one coat. Internal coherence asks whether the intervention fit with the organisation's own other work, policies, and commitments. External coherence asks whether it fit with what everyone else other agencies, government systems, other donors was doing in the same space at the same time.
Did this project duplicate, contradict, or quietly undercut something else? Did it plug into existing national systems, or build a parallel structure that collapses the day funding ends? Coherence is where donors look for evidence that you understood your project was never operating alone and where they look for honesty about friction, because real coordination is rarely as smooth as proposals suggest.
For example: a health project that set up its own community data system, running separately from the national health information system, may have produced clean data for the donor. But if it duplicated government effort, confused health workers reporting the same numbers twice, and left nothing behind at close, that's a coherence failure, even if every indicator was met.
How to handle it: name the other actors specifically. Map where the project aligned with government and other donors, and where it diverged. Be honest about the partnerships that worked and the ones that created friction.
Effectiveness"Did the project achieve its objectives?"
This is the most familiar criterion, and the one most likely to be answered with a wall of indicator numbers: targets met, targets missed, percentages against the logframe. It's the section evaluators feel most comfortable writing, which is precisely why it's so often the weakest.
Not whether you hit the targets, but whether you understand why. A donor can read an indicator tracking table themselves. What they're paying an evaluator for is the causal story which activities actually drove the results, which assumptions held and which broke, and which apparent successes were really just favourable conditions the project happened to operate in.
For example: two districts both hit their target for trained health workers. In one, training changed practice because supervisors reinforced it; in the other, trained staff transferred out within months and the knowledge left with them. The indicator looks identical. An evaluation reporting only the percentage misses the most useful finding in the project.
How to handle it: lead with the explanation, not the table. For each major result, ask what actually caused it, and be specific about the conditions that made success possible or prevented it. The numbers establish what happened; your value is in credibly explaining why.
Efficiency"Was it worth what it cost?"
Efficiency covers both money and time: were resources converted into results economically, and did things happen when they were supposed to? This is the criterion teams most often dodge, usually because the financial data is messy, arrived late, or was never properly tracked in the first place.
Increasingly, value for money and donors are considerably more cost-conscious now than they were even five years ago. A vague efficiency section makes a donor assume the worst. They're looking for evidence that you engaged seriously with cost relative to results, with the cost of delays, and with the trade-offs the project made under real-world pressure.
For example: a project that delivered all its outputs but took eighteen months longer than planned has an efficiency story to tell, whether the evaluator tells it or not. If the delay came from a slow procurement process that could have been anticipated, that's a finding. Omitting it doesn't make it invisible to a donor watching the disbursement schedule.
How to handle it: you don't need a full cost-effectiveness analysis. You need to actually engage with cost per outcome where you can estimate it, with delays and what they cost, and with the trade-offs made when budgets tightened. Honesty about a delay reads far better than silence about it.
Impact"What difference did it really make?"
Impact asks about the higher-order, longer-term change; the difference the project made beyond its immediate outputs and outcomes. It's the most ambitious criterion and the one where evaluators most often overreach.
Two things weaker evaluations routinely skip. First, attribution versus contribution: are you claiming the project caused a change, and can you defend that, or is the honest framing that it contributed alongside other forces? Overclaiming attribution is the fastest way to lose a sophisticated reader.
Second, unintended effects, including negative ones. An impact section that reports only good news nobody planned for is not a serious impact section.
For example: a girls' education project might report rising enrolment as its impact. A credible evaluation asks how much of that rise the project can actually claim, given that a new school feeding programme and a good harvest year also pushed enrolment up and whether there were unintended effects, such as longer walking distances creating new safety risks.
How to handle it: be disciplined about causal language. Use "contributed to" unless you can genuinely defend "caused." Actively look for unintended effects, good and bad. Donors trust evaluators who surface the difficult findings, because it signals the favourable ones were scrutinised just as hard.
Sustainability"Will the benefits outlast the funding?"
Sustainability asks what continues once the funding stops — the benefits, the capacities, the systems. Almost every project writes itself a confident sustainability plan; far fewer can show that the plan was ever real.
The honest answer, not the hopeful one. They're reading for evidence; handed-over assets that are actually being used, local ownership that existed well before the closing workshop, and financing that doesn't quietly depend on a follow-on grant. A sustainability section built on optimism rather than evidence fools no one who has watched projects end.
For example: a water project that built boreholes and trained a maintenance committee has a real question: is there a functioning, funded mechanism to pay for repairs when a pump breaks in two years? If no one collected user fees and no spare-parts supply chain exists, the benefits are fragile and saying so is more useful than an optimistic paragraph everyone discounts.
How to handle it: look for evidence of what will actually survive, not statements of intent. Where sustainability is genuinely fragile, say so plainly and explain what would be needed to secure it.
The criteria are a conversation, not a checklist
Read together, the six criteria are really one extended question a donor is asking: did this project understand its context, fit into it, work, justify its cost, change something real, and leave something behind? When you organise a report mechanically, six sections, findings dropped under each, you answer the labels but miss the conversation. When you write to the subtext, the same evidence does far more work, and the report reads like it was produced by someone who has actually run evaluations rather than someone completing a template.
That distinction is most of what separates a report that merely satisfies a terms of reference from one a donor remembers and acts on. The criteria reward evaluators who treat them as a way of thinking, not a way of formatting.
Planning an endline or final evaluation?
At M & E Lab Consultancy Limited (MELAB), designing and leading credible OECD-DAC evaluations is core to what we do across health, WASH, protection, livelihoods, education, and governance programming in East and Southern Africa. If you want an evaluation that reads as credible to the people funding it, let's talk.
📞 +255 747 454 795 (Call / WhatsApp) · ✉️ info@melabconsulting.com · 🌐 melabconsulting.com
