Doing the Right Things? It Depends. Why Relevance Requires More Than Alignment with Needs and Priorities
Can an intervention address a genuine development need, align with government priorities, follow a well-established development approach, and still be only partially relevant?
The answer should be yes.
Yet in many development evaluations, relevance remains one of the easiest OECD-DAC criteria on which programmes receive high ratings. The reasoning is often straightforward: the problem exists, beneficiaries confirm that it matters, the intervention is aligned with national policy, and its objectives correspond to donor or institutional priorities. The conclusion follows: the programme is highly relevant.
But this reasoning may be missing something important.
The OECD-DAC defines relevance as the extent to which an intervention's objectives and design respond to beneficiary, country, global and institutional needs, policies and priorities, and continue to do so as circumstances change. Importantly, OECD's subsequent guidance identifies not only needs and policy alignment, but also context, quality of design and responsiveness over time as dimensions of relevance. It explicitly brings the intervention's theory of change, theory of action and modus operandi into consideration.
This has a significant implication:
The existence of a relevant problem does not automatically establish the relevance of the chosen response.
Development problems rarely prescribe their own solutions
Consider an area characterised by persistent rural poverty, low agricultural productivity, weak market access and highly vulnerable households.
A livelihoods programme might interpret the central problem as insufficient household assets and capabilities. Its response could include productive assets, skills development, access to finance, producer organisation and direct market linkages.
A Market Systems Development (MSD) programme may look at the same area and reach a different diagnosis. It may see poor household outcomes as symptoms of deeper systemic constraints: weak service markets, poor incentives for private-sector actors, information asymmetries, restrictive rules, weak coordination and market functions that exclude poorer producers.
Both perspectives can be legitimate.
Both may address real problems.
And both are based on different assumptions about where change needs to occur and how sustainable development happens.
This matters enormously for evaluation.
If evaluators simply establish that farmers need higher incomes and they lack access to inputs and output markets, and that an MSD programme seeks to improve their incomes, relevance appears straightforward. But the real evaluative question should go further:
Was the programme's understanding of why farmers had low incomes sufficiently complete, and was its chosen intervention approach appropriate to those constraints?
That moves relevance from problem alignment to problem–response fit.
Relevance is theory-laden
Development interventions are never theoretically neutral.
An MSD programme rests, explicitly or implicitly, on ideas about systems, incentives, market functions, facilitation, crowding-in and sustainable behaviour change.
A livelihoods programme may be grounded more strongly in household capabilities, vulnerability reduction, asset accumulation and resilience.
A social protection programme, graduation programme, governance intervention or rights-based programme will bring still other assumptions about causes, agency and appropriate forms of intervention.
Consequently, what constitutes an “appropriate response” cannot be determined independently of the programme's theory of change.
This does not mean relevance is simply subjective.
The evaluator should not replace the programme's development philosophy with their own:
“We employ systemic approaches; therefore, direct livelihood support is irrelevant.”
Nor:
“Poor households needed assets, therefore MSD was inappropriate.”
Both would be weak evaluative reasoning.
Instead, relevance should test whether the reasoning connecting need, diagnosis, causal assumptions and intervention choice is defensible against the available evidence.
A useful way of thinking about this is:
Relevance concerns the integrity of this chain.
What if the problem exists at more than one level?
This becomes particularly important where development problems are multidimensional.
Suppose an evaluation finds that poor rural households face two categories of binding constraints.
At the household level, they have few productive assets, little working capital, high indebtedness and extremely limited capacity to absorb risk.
At the market-system level, they face weak input and service markets, limited relationships with buyers, poor information and incentives that discourage firms from serving poorer producers.
An MSD intervention may be highly relevant to the second set of constraints.
But what happens if households lack the minimum capabilities required to take advantage of the opportunities created by market-system change?
Improved commercial services may become available, but poorer farmers cannot afford them.
Buyers may offer new market opportunities, but some producers lack the assets needed to increase production.
Financial services may expand, while households considered too risky remain excluded.
The MSD intervention has not suddenly become irrelevant. It is addressing genuine structural constraints.
But an MSD-only strategy may be insufficiently responsive to the full structure of the problem.
That distinction is important.
An intervention can be relevant but insufficient
Evaluation practice would benefit from explicitly recognising this category.
Rather than forcing a binary choice between:
Relevant or Not relevant
we should be comfortable finding that:
The chosen approach was relevant to an important dimension of the problem, but the overall design was insufficiently responsive to other binding constraints necessary for the intended causal pathway to operate.
This is quite different from an effectiveness finding.
The evaluator is not yet asking whether the intervention successfully changed the market system. That belongs principally under effectiveness.
The relevance question is whether, given what was known or reasonably knowable about the target population and context, the intervention architecture constituted an adequate response to the problem it intended to address.
The missing intervention test
This also requires caution.
Every programme has boundaries. A food security programme cannot be criticised for failing to solve every education, health, infrastructure and governance problem in its operating area.
The relevance standard therefore cannot be:
Did the programme address every need?
A more useful question is:
Did it address or adequately account for the constraints necessary for its own intended causal pathway to work?
This provides an important test for evaluators considering a “missing” intervention.
Suppose livelihood support would simply have made an MSD programme somewhat better. Its absence should not necessarily reduce the relevance rating.
But suppose the programme expects vulnerable households to respond to improved market opportunities, while those households lack the minimum assets, liquidity or resilience needed to participate.
Now the missing livelihood component is no longer merely desirable.
It may represent a missing causal condition.
That is a legitimate relevance finding.
Relevance may differ across beneficiary groups
The same intervention may also have different levels of relevance for different segments of the population.
An MSD approach might be highly appropriate for commercially oriented smallholders who have some productive assets and capacity to invest.
It might be only partially responsive to poorer farmers facing severe liquidity and asset constraints.
And it may have limited direct relevance for landless or extremely vulnerable households unless specific market opportunities exist for them.
An aggregate statement that “the intervention was relevant to beneficiary needs” can therefore conceal substantial variation.
This is consistent with OECD-DAC's emphasis on recognising differences and trade-offs between needs and priorities, and on placing beneficiaries centrally in relevance assessments.
Relevance should consequently ask not only:
Relevant to what?
but also:
Relevant to whom?
and:
Relevant for addressing which constraint?
Sometimes the answer is not either/or, but sequencing and layering
Complex development problems may require intervention at different levels and different points in time. For some vulnerable populations, an appropriate pathway might look more like:
Temporary livelihood support and systemic market facilitation need not necessarily represent competing development philosophies.
They may perform different functions.
Livelihood interventions may enable extremely vulnerable households to reach a threshold at which they can participate in markets.
Market-system interventions may then address the structural conditions necessary for those gains to become scalable and sustainable.
In other contexts, both may operate simultaneously but target different beneficiary segments.
The relevance question therefore becomes:
Did the programme differentiate, sequence or combine interventions in a manner consistent with the different constraints facing different groups?
This is more useful than asking whether MSD or livelihoods programming was the “correct” approach.
And sometimes complementarity belongs under coherence
There is another important qualification.
No single programme needs to address every necessary constraint itself.
An MSD programme may legitimately concentrate on market-system functions if complementary programmes provide social protection, humanitarian support, livelihood assets or other services needed by highly vulnerable households. In such circumstances, the evaluator should avoid penalising the programme merely for maintaining an appropriate intervention boundary.
The question increasingly shifts towards the OECD-DAC criterion of coherence: whether interventions complement one another, create synergies, avoid duplication and collectively provide a sensible response.
So the distinction matters:
Relevance asks: Is this intervention appropriately designed in relation to the needs, constraints and context it seeks to address?
Coherence asks: How well does it fit with other interventions addressing related parts of the problem?
A good evaluation needs both lenses.
From “doing the right things” to defensible intervention logic
OECD-DAC summarises relevance through the deceptively simple question:
“Is the intervention doing the right things?”
Perhaps the question needs to be operationalised more carefully.
There may be no universally agreed “right thing” independent of context, values, development theory and causal assumptions.
A stronger relevance enquiry might therefore ask:
Given the needs and priorities, the diagnosed causes of those needs, the capacities and characteristics of different population groups, the operating context and plausible alternative responses, was there a defensible rationale for choosing these objectives and this intervention design?
This suggests four layers for analysing relevance:
Relevance of the problem: Does the identified problem or need genuinely matter?
Relevance of the diagnosis: Does the intervention adequately understand the important causes and constraints producing the problem?
Relevance of the response: Is the chosen intervention approach plausible and appropriate given that diagnosis and context?
Continued relevance: Do those assumptions and intervention choices remain appropriate as needs and circumstances change?
This is not about unnecessarily complicating evaluations. It is about avoiding a much simpler analytical error:
Need exists → program addresses need → program is highly relevant.
Relevance is ultimately about problem–response fit
Development evaluation should not become an arbiter between competing schools of development practice.
An MSD programme should not automatically be considered more relevant because it seeks systemic change. A livelihood intervention should not automatically be considered more relevant because it responds directly to immediate household needs.
Both approaches can be highly relevant.
Both can also be poorly matched to a particular problem.
And sometimes each addresses only one part of a problem that operates simultaneously at household, community, market and institutional levels.
The evaluator's task is therefore not to determine which development doctrine is appropriate.
It is to examine whether the intervention architecture fits the structure of the problem.
Perhaps this is the most useful way to understand relevance:
Relevance is not simply about whether an intervention addresses an important problem. It is about whether the reasoning that connects the problem to the chosen response is credible, sufficiently complete and appropriate to the people and context concerned.
Once viewed this way, relevance becomes much more than a routine check on beneficiary needs and policy alignment.
It becomes one of the most intellectually demanding, and potentially most useful judgments an evaluation can make.
References
OECD (2019). Better Criteria for Better Evaluation: Revised Evaluation Criteria Definitions and Principles for Use.
OECD (2021). Applying Evaluation Criteria Thoughtfully.
Comments