Plenty of data, few decisions: how to design MEL that guides action
Published 20.10.2026/Monitoring/20 min to read
In 30 seconds
Most MEL systems are built to report, not to decide. The two goals look alike but call for different indicators, frequencies and governance.
Three design choices explain why the system goes quiet when it matters: it starts from the logframe, data arrives too late or too aggregated, and nobody owns the step from finding to decision.
Four principles fix it: start from the decisions, write decision milestones into the theory of change, set data frequency by the pace of the decision, and appoint an owner for interpretation.
The dashboard is up to date. The quarterly report went out on time. Every indicator is filled in, to the decimal.
DashboardUp to date
Quarterly reportSent on time
IndicatorsFilled in, to the decimal
Decision: reorient a component?The system has nothing to say
Then comes the day the programme manager has to decide: reorient a component, shift budget from one activity to another, adjust the implementation strategy. And the monitoring system, so diligent until now, has nothing to say. The decision gets made elsewhere, on other grounds.
This is not bad luck. It is a design choice that is rarely made out loud: most monitoring, evaluation and learning (MEL) systems are built to report, not to decide. The two goals look alike, but they call for different indicators, different collection frequencies and different governance of information. Merge them, and you get a system that is excellent at proving what happened and silent on what to do next.
Same system, two designs
Indicators
Activities carried out, outputs delivered, targets met or approached.
Collection frequency
Annual, consolidated at national level.
Governance of information
The system stops the moment the report is published.
What it produces
Proof of what happened.
Excellent at proving what happened. Silent on what to do next.
Indicators
The data a manager would turn to when deciding whether a component is on track or needs to change course.
Collection frequency
At the pace of the decision, broken down at the level where it acts.
Governance of information
Someone owns the step from finding to decision.
What it produces
A decision, at the right time, for the right person.
Informs a specific decision, at the right time, for the right person.
Why your MEL system goes quiet when it matters
Three design choices explain the silence.
01Starting point
It starts with the logical framework, rather than with the questions project managers actually need to answer.
Indicators are selected to meet donor reporting requirements: activities were carried out, outputs were delivered, and targets were met or approached. Yet these are rarely the indicators a project manager would turn to when deciding, halfway through a project, whether a component is on track or needs to change course.
02Timing
The data arrives too late, or too aggregated.
An annual indicator, consolidated at national level, cannot help adjust a municipal intervention that is already under way. Monitoring has to run at the pace of the decision it is meant to inform. Otherwise it is just an archive of the past.
03Ownership
Nobody owns the step between a finding and a decision.
Many systems stop the moment the report is published. No one is explicitly asked to say: “here is what this means, and here is what we do about it.” Without that loop, even excellent data stays inert.
The good news: all three are design problems, so all three can be designed out.
Four principles for MEL that guides action
Evaluation standards, the OECD DAC criteria first among them, insist on one point: relevance, effectiveness and efficiency are not boxes to tick. They only produce a useful evaluation when applied thoughtfully, in context. The same logic holds for monitoring. A system is not good because it tracks many indicators. It is good because it informs a specific decision, at the right time, for the right person.
Four principles follow.
1
Start from the decisions, not the indicators.
Instead ofBuilding the monitoring matrix first, then hoping the indicators will be used.
Do thisList the decisions of the next twelve months, then work backwards to the data.
Before building the monitoring matrix, list the decisions the management team will face over the next twelve months: reallocating a budget, suspending an activity, scaling up a pilot. Then work backwards to the data that would inform each one. This reversal keeps you from collecting indicators nobody will ever use.
2
Write decision milestones into the theory of change.
Instead ofA chain of results with an annual target on its own.
Do thisSay when, and on which thresholds, a decision will be taken.
A theory of change should not only describe a chain of results. It should say when, and on which thresholds, a decision will be taken. “If the adoption rate is still below X% at mid-term, the team revises its awareness strategy” is a decision milestone. An annual target on its own is not.
Try it: a decision milestone in action (illustrative)
IF adoption rate at mid-term <40%THEN revise the awareness strategy
threshold 40%
Below 40% at mid-term: milestone triggered. The team revises its awareness strategy.
3
Set data frequency by the pace of the decision, not the donor’s calendar.
Instead ofOne frequency for everything, dictated by the reporting calendar.
Do thisMonthly committee, monthly data, at the level where it acts. The annual report stays annual.
A monthly steering committee needs monthly data, broken down at the level where it acts. A consolidated annual report can stay annual. These are two products with two audiences, and confusing them is a classic source of waste.
4
Appoint an owner for interpretation, not just for collection.
Instead ofData that is collected, published and left to speak for itself.
Do thisA recurring moment where someone has the explicit mandate to say what the figures mean.
Data does not speak for itself. Build in a recurring moment, such as a steering review or a monthly check-in with field teams, where someone has the explicit mandate to say what the figures mean and to propose a decision. Without that role, analysis stays a technical exercise, cut off from action.
One caveat: reporting still matters
None of this means giving up on reporting to donors and stakeholders. Both functions have to coexist, but they should be designed separately from day one. Asking a single system to serve institutional reporting and day-to-day management at once is often the root cause of systems that produce a lot of data and few decisions.
Institutional reporting
Audience
Donors and stakeholders
Frequency
Consolidated, annual
What it protects
The overall picture that accountability and institutional learning depend on
Two products, designed separately from day one
Day-to-day management
Audience
The management team and field teams
Frequency
At the pace of the decision, broken down at the level where it acts
What it protects
The ability to adjust an intervention while it is under way
The opposite mistake is just as costly. Cut monitoring down to a handful of management indicators and you lose the overall picture that institutional learning and accountability depend on. The goal is a design that connects the two clearly, with distinct products, frequencies and audiences.
Where to start
A MEL system that guides action starts before the first data collection. It starts with a conversation with decision-makers about the trade-offs they will have to make, and with a theory of change that records those trade-offs as explicit milestones. Indicators, collection tools and reporting frequency follow from there, not the other way round.
Step 1
A conversation with decision-makers about the trade-offs they will have to make.
Step 2
A theory of change that records those trade-offs as explicit milestones.
Step 3
Indicators, collection tools and reporting frequency follow from there.
At Artemis, this is how we approach it: anchor MEL in a theory of change from the design stage, define indicators that serve decisions and not only reporting, and build in regular reviews where the team analyses, decides, learns and adjusts. A good MEL system does not just watch. It informs, it questions, it adjusts.
Theory of change
Anchored from the design stage, with the trade-offs recorded as explicit milestones.
Useful indicators
Defined to serve decisions, not only reporting.
Regular reviews
Moments where the team analyses, decides, learns and adjusts.
A good MEL system does not just watch. It informs, it questions, it adjusts.
Our infographic: integrating M&E early changes the policy trajectory.
A question for your own projects: when do you really bring monitoring and evaluation in?
Then your theory of change can carry decision milestones from day one. The next article in this series, on building a theory of change that works as a real compass, is written for you.
You are in good company: most systems are retrofitted. The four principles above are a way to start the retrofit from the decisions, not from the indicators.
1Series
Rethinking MEL
This article opens our series “Rethinking MEL”.
Next: how to build a theory of change that works as a real compass, not a methodological exercise in compliance.
Our SJIAOS paper on what really drives energy consumption across the EU-27.
Manage Consent
Artemis uses cookies to ensure our site works properly and to improve your experience. Learn more in our cookie policy.
Functional
Always active
The technical storage or access is strictly necessary for the legitimate purpose of enabling the use of a specific service explicitly requested by the subscriber or user, or for the sole purpose of carrying out the transmission of a communication over an electronic communications network.
Preferences
The technical storage or access is necessary for the legitimate purpose of storing preferences that are not requested by the subscriber or user.
Statistics
The technical storage or access that is used exclusively for statistical purposes.The technical storage or access that is used exclusively for anonymous statistical purposes. Without a subpoena, voluntary compliance on the part of your Internet Service Provider, or additional records from a third party, information stored or retrieved for this purpose alone cannot usually be used to identify you.
Marketing
The technical storage or access is required to create user profiles to send advertising, or to track the user on a website or across several websites for similar marketing purposes.