The KPI Nobody Tracks: How to Measure Learning (Not Just Performance) in Your MEAL System

When I help organizations audit their MEAL systems, I almost always find the same pattern. The indicator matrix is full. The IPTT has columns for baseline, targets, and actuals. The quarterly reports go out on time. And yet, when I ask "what did your team learn last quarter, and what did you change because of it?", the room goes quiet. Performance measurement? Covered. Learning measurement? Almost never. This is the KPI nobody tracks, and I think it's the most important gap in most MEAL systems today. Why Performance KPIs Dominate (And Learning Gets Left Out) There is nothing wrong with tracking performance. Donors expect it. Program managers need it. But there is a real problem with only tracking performance: a number that describes what happened tells you almost nothing about whether your organization is getting smarter about how to make things happen better. The data on KPI quality in general is uncomfortable. Research from ClearPoint Strategy, which analyzed nearly 7,000 education KPIs across schools and colleges, found that only 30 out of every 100 KPIs added to a strategic plan were still being updated a year later, and 28 never received a single data point [1]. That means the majority of what organizations promise to measure simply... disappears. No one owns it, no one updates it, no one acts on it. In humanitarian and development MEAL systems, the dynamic is familiar: outputs and participation numbers survive because they feed directly into donor reports. Learning questions, "Did staff apply what they learned in the training?" "Did the community feedback change our approach?", rarely make it into a formal indicator, so they drift. ⚠️ Warning: A KPI without an owner is a KPI with no future. As the ClearPoint data shows, three of every four college KPIs belong to no one [1]. Ownership is not a bureaucratic detail, it is what transforms a measurement intention into an actual data point. The deeper issue is a systems one. As Lean Six Sigma practitioners have long argued, the danger of KPIs is not that we measure, it is that we measure without understanding the system that produces the results [4]. Tracking outputs tells you the system ran. Tracking learning tells you whether the system is capable of improving itself. What "Measuring Learning" Actually Means Before I get into the how, it helps to be precise about what we mean. Learning in a MEAL context is not the same as training completion. Learning means: did new information or experience change what a team believes, decides, or does? There are three levels worth distinguishing: Information uptake, Did the team receive and understand findings? (e.g., survey results shared, feedback loops closed) Reflection and sense-making, Did the team discuss findings and draw conclusions? (e.g., learning review held, key takeaways documented) Adaptive action, Did something change as a result? (e.g., activity design updated, indicator revised, process improved) Most MEAL systems, if they touch learning at all, stop at level one. The data gets shared. Whether it changed anything is never tracked. 💡 Tip: The cleanest test of a learning KPI is this: if the number moved, would you actually do something differently? If the answer is no, the indicator is measuring a ritual, not learning. What Good Learning KPIs Look Like Good KPIs have a specific structure: a clear name, a unit of measurement, a named data source, a target, an intervention threshold, and an owner [5]. Learning KPIs are no exception, they just apply that structure to learning processes rather than program outputs. Here are practical examples across the three levels: Learning Level Example Indicator Unit Data Source Owner Information uptake % of staff who received and confirmed review of monthly monitoring data Percent Distribution log + signed acknowledgment MEL Officer Reflection Number of documented learning sessions held per quarter Count Learning log / meeting minutes Program Manager Adaptive action % of flagged recommendations with documented follow-up action within 60 days Percent Recommendation tracker MEL Coordinator Adaptive action Number of program adjustments made based on beneficiary feedback in the period Count Feedback register + action log Deputy Director Notice that none of these require a sophisticated new system. Most of the data sources are things that should already exist: meeting minutes, feedback registers, action logs. The gap is usually not a data gap, it is a decision gap. No one decided to treat learning as something worth measuring. 📝 Note: "Number of learning sessions held" is an output indicator for learning, not a learning indicator itself. Use it as a proxy only when paired with at least one adaptive-action indicator that tracks whether those sessions produced change. The Most Common Pitfall: Confusing Activity With Learning I see this constantly. A MEAL plan will include indicators like "number of staff trained" or "number of lessons-learned workshops conducted." These are outputs. They describe that a learning activity happened, not that learning occurred. This matters for the same reason that measuring training attendance tells you nothing about behavior change. An organization can hold twelve lessons-learned workshops per year and make zero program adjustments. The activity happened. The learning did not. ⚠️ Warning: If your only learning-related indicator is a count of workshops or reviews conducted, you are measuring scheduling, not learning. Add at least one indicator that tracks what changed because of those sessions. The fix is to pair every process indicator with an outcome indicator one level up. If you track "learning reviews conducted," also track "percentage of reviews that produced a documented recommendation." If you track "recommendations documented," also track "percentage of recommendations acted on within the period." Building Learning KPIs Into Your MEAL System: A Practical Approach Here is how I would approach embedding learning measurement into an existing MEAL system without starting over: Audit your current indicator matrix. Identify which indicators (if any) relate to learning processes or adaptive management. Most matrices will have none. Pick one indicator per learning level. Start with information uptake, reflection, and adaptive action. Three indicators is enough to reveal the shape of your learning culture. Assign an owner to each. A learning indicator with no named owner will be abandoned within two reporting cycles. This is not a prediction, it is what the data on KPI survival consistently shows [1]. Define your data source before you finalize the indicator. If you cannot name where the data will come from and who will collect it, the indicator is aspirational, not operational [5]. Report on learning indicators in the same table as performance indicators. Separating them into a "learning annex" is how they get treated as optional. Put them in the main IPTT. Set a meaningful target. "100% of recommendations acted on" is probably unrealistic. "60% of flagged recommendations with documented follow-up within 60 days" is both honest and actionable. 💡 Tip: Consider adding a standing agenda item to your monthly program review: "What did monitoring data tell us this month, and what are we changing?" Documenting that conversation takes five minutes and becomes the data source for your reflection-level indicator. The Signal You Are Probably Missing Here is the real cost of not tracking learning: programs keep making the same preventable mistakes because no system is designed to stop them. A community feedback loop surfaces a problem in month three. It gets logged. It gets reported. And in month nine, the same problem surfaces again, because the feedback never triggered a change, and no one was measuring whether it had. Effective KPIs help organizations "identify early warning signs before they become major issues" and "drive continuous improvement through data-driven decisions" [3]. But that only works if the system is designed to improve itself, not just describe itself. Learning KPIs are how you close that loop. If you would like help designing a learning-focused indicator set or embedding learning KPIs into an existing MEAL framework, that is exactly the kind of work I do at vera.ignex.io. You can try a live session at no cost and see whether it fits your team's needs. The programs that learn fastest will always outperform the programs that simply report fastest. Building that into your measurement system is one of the highest-leverage improvements you can make. Follow Vera for more on MEL & project management: LinkedIn · Instagram · Facebook · X