A monitoring and evaluation system that outlives the project

In most development programmes, monitoring and evaluation (M&E) is experienced as a donor obligation: a reporting machine cut off from the action, whose real function ends on the last page of the final report. That is a waste, and it is measurable. A well-designed M&E system is not an administrative cost: it is the navigation instrument of a public policy, the only way to know whether the money spent is delivering the promised result, and to correct course before it is too late. So the problem is not that we evaluate too much, but that we evaluate badly, and above all that we build systems tailored to the donor rather than to the country. The decisive question is not "what can we measure?", it is "will this system still be of any use the day the funding stops?". For most, the answer is no. Here is why, and six principles to turn that answer into yes.

The diagnosis: systems built for the report, not for the decision

This concern is not a practitioner's impression: it shows up in the evaluators' own data. The World Bank's Independent Evaluation Group (IEG), which rates the quality of monitoring and evaluation for every project at closure, reveals a stubborn reality: for years, only a minority of projects reached an M&E quality rated "substantial" or "high". The share of well-rated projects has risen, from around 57 % in 2020 to 64 % in 2021, before settling back to 63 % in 2022, but the message is the same: more than one project in three, in the institution that invests more than any other in evaluation, exits its cycle with a measurement system judged weak. If that holds true where resources and expertise are densest, the gap is mechanically wider for West African national programmes, where M&E is often the first line cut when budgets tighten.

The right framework already exists: the OECD's six criteria

Before building a system, one must know what evaluating means. The global reference is stable and shared: the six criteria of the OECD Development Assistance Committee (DAC), revised in 2019 and now numbering six with the addition of coherence. They come down to six simple questions. Relevance: does the intervention meet a real need? Coherence: does it fit with other policies rather than duplicate them? Effectiveness: does it achieve its objectives? Efficiency: at what cost, and could more have been done with the same means? Impact: what lasting effects, intended and unintended, on people? And sustainability: will the benefits outlive the funding? This last criterion is, for our subject, the most revealing, because it applies to the M&E system itself as much as to the project it measures: a system that dies with the donor fails on precisely the criterion it was meant to monitor.

This framework has an often-overlooked practical virtue: the OECD itself recalls that these criteria are not merely boxes to tick at the end, but questions to ask from the design stage, during implementation and throughout monitoring. In other words, the good M&E system is not the one that issues the verdict at the end; it is the one that asks these six questions continuously, while the action unfolds and there is still time to act.

Why so many systems die with the project

If we look at the M&E systems that collapse at a programme's closure, the same causes recur. They are not technical but structural, and they all trace back to one original sin: the system was designed for the donor, not for the country.

• Indicator overload. A logical framework weighed down with dozens of indicators looks rigorous; in reality it becomes impossible to fill in faithfully, and half the cells stay empty or approximate. A system you cannot maintain is not monitored, it is merely endured.
• Dependence on external expertise. When data collection, analysis and dashboards rest on an international consultant or a proprietary tool no one on site has mastered, the system leaves with them the day the contract ends.
• No institutional anchor. A system housed in a temporary project unit, rather than in a permanent ministerial directorate, has no heir. When the unit closes, the data and the skills scatter.
• Funding tied to the project budget alone. If the M&E line disappears with the funding, no one pays for collection, database maintenance or report production. The system does not die for lack of usefulness, it dies for lack of an operating budget.

None of these causes is inevitable. All can be corrected at the design stage, provided the right question is asked from the start. The six principles that follow are precisely the answer to each of these breaking points.

A good monitoring and evaluation system is not there to render accounts: it is there to make better decisions, sooner. And it keeps serving them long after the donor has gone.

Six principles for a system that lasts

The principles below are not yet another recipe. They answer, one by one, the causes of death of M&E systems, and they reinforce one another: the first disciplines demand, the next make it sustainable, and the last guarantees that it survives.

1. Start from decisions, not from indicators

The first question to ask is never "what can we measure?", but "which decisions must we inform, and who takes them?". An indicator that informs no decision is a pure cost: it must be collected, entered, checked, reported, and it changes nothing in the action. Building a useful M&E system therefore begins by mapping the real decisions, at every level, from the minister arbitrating a budget to the field supervisor redirecting a round, then working back from each decision to the minimum information that informs it. This inversion of the usual logic, which starts from available data rather than from the choices to be made, is what separates a living dashboard from a dead database. It rests on a proven tool, the theory of change, which makes the "activities, outputs, outcomes, impact" chain explicit and lets each indicator be placed against the link it serves.

2. Tighten rather than accumulate

The opposite reflex is everywhere: out of caution, indicators pile up, as if rigour were measured by their number. That is the founding error. A results framework overloaded with fifty indicators ends up ignored, because no team has the means to fill them all in properly. Better a small number of robust indicators, faithfully and reliably tracked, than a forest of half-filled indicators that gives the illusion of measurement without its reliability. Parsimony is not a surrender of rigour, it is its condition: an indicator you actually track is worth infinitely more than ten you merely declare. The good test is simple: for each indicator, if you cannot name the decision it informs nor the person who will fill it in each quarter, it should not be in the framework.

3. Digitise collection, automate reporting

Paper-based collection imposes a delay and a data-entry cost that condemn M&E to always looking at the past. Digital collection changes the nature of the tool: data flow back in near real time, quality controls are built into the source (an enumerator cannot enter an age of 200 years nor skip a mandatory question), and reporting is generated automatically as readable dashboards, rather than as frozen reports produced three months after the fact. This shift is no technological luxury: it is what turns an accountability system, looking backward, into a steering instrument, looking toward the decision.

• Data available in near real time, not months after collection.
• Consistency and completeness checks built into the form, at the data source.
• Dashboards and reporting generated automatically, readable by decision-makers and field teams alike.
• Traceability and geolocation that reveal not only how many, but where, and therefore where to act.

4. Train, transfer, foster ownership

This is where sustainability is decided, and this is where most projects fail. A system is durable only if national teams can run it without external assistance. Training therefore cannot be a module wrapped up in the project's final weeks, when technical-assistance funding runs dry: it is a continuous skills transfer, built in from the design stage, explicitly targeting the staff who will keep the system alive after the donor has gone. The literature on national ownership converges on this point: the durability of a measurement system rests on a few concrete dimensions, the partnership, the commitment and responsibility of national actors, their capacities and their accountability. None of these dimensions can be bought: they are built over the long term, by having others do rather than doing in their place. A system delivered turnkey but not owned is a system already doomed.

5. Close the loop back to action

An M&E system that collects without reporting back is a well without a pump. Data must return to those who act, in a readable form and at a regular pace, failing which M&E degenerates into an administrative ritual disconnected from the very action it was meant to serve. This feedback loop, from data to decision and then to the adjustment of action, is what distinguishes a living system from an archive. It calls for reporting calibrated for each audience: a synthetic dashboard for the decision-maker, operational indicators for the project manager, geolocated feedback for the field supervisor. When the loop works, M&E stops being seen as an imposed chore and becomes a service the teams ask for, because it helps them do their job better. That, incidentally, is the best guarantee of durability there is: a system that is used does not die.

6. Plan for life after the project from day one

The last principle is in fact the first to address, from the very start: who will maintain this system when the funding stops, in which institution will it be housed, and with what operating budget? The answer to this triple question determines success more surely than any indicator, and it must be written down in black and white in the design, not deferred to the final review. Concretely, that means anchoring the system in a permanent directorate rather than a temporary project unit, favouring open, documented tools that national teams can maintain, and securing an operating budget line, modest but recurrent, distinct from project funding. An M&E system becomes durable only when its survival no longer depends on the project that gave birth to it. This is exactly the approach CRAD takes when it designs, installs and trains partners on their monitoring and evaluation systems: not to deliver a tool, but to transfer a capacity.

The cost of inaction: a dead system blinds the entire investment

Monitoring and evaluation represents a modest share of a programme's budget. The standards of major agencies converge around 3 to 10 % of the total budget, the most cited order of magnitude being 5 to 10 % depending on programme complexity, with several donors such as UNDP or USAID sitting closer to 3 to 5 %. This spending is among the most profitable in the programme, because it conditions the value of everything else: it is what tells you whether the remaining 90 to 97 % are producing the expected effect, and lets you correct in time if they are not. Cutting M&E to save a few budget points amounts to flying a plane by unplugging the instruments to lighten the aircraft. The real cost is not that of the measurement system, it is the cost of its absence: a programme that proceeds without reliable data spends its entire envelope blind, and will never know whether it paid off.

The CRAD angle: transfer a capacity, not deliver a tool

The conviction guiding our work comes down to one sentence: a monitoring and evaluation system has value only if it outlives the one who installed it. That is why we design our partners' M&E systems around national ownership, not around our own presence. In practice, that translates into demanding choices that cut against the easy path: tightening the indicator framework until it is genuinely sustainable by the teams in place, rather than impressing the donor with its exhaustiveness; building dashboards the decision-maker consults of their own accord, because they answer their trade-offs; and above all training, continuously and in the field, the staff who will keep the system alive once the project has closed. This is the approach we bring to our missions setting up M&E systems and delivering training, in public health as in agricultural value chains or energy.

This philosophy of ownership is part of a broader movement: national evaluation networks and regional capacity-building centres, such as the CLEAR centres, show that the path to durability runs through systems genuinely owned by countries, not grafted on from outside. A French-speaking research firm rooted in West Africa has a particular role to play here: bridging the methodological demands of donors and the reality of the institutions that will have to run the system alone, once the funding has gone. It is in that gap, between the client's requirement and the country's capacity, that the durability of a measurement system is won or lost.

Ultimately, the durability of an M&E system is not one more technical question, it is the question. A system that dies with the project has not merely failed to last: it has failed on the very criterion of sustainability it was meant to monitor in others. The six principles set out here demand neither out-of-reach technology nor an outsized budget. They demand a discipline of design and a change of gaze: stop building systems for the final report, and start building systems for the decision, for the country, and for the after.

← All insights