Designing a MEL plan that survives the field
From realistic indicators to collection tools: how to build a monitoring & evaluation system that teams actually use — and that survives field realities, from patchy connectivity to staff turnover.
Resources
Articles, courses and videos on monitoring & evaluation, GIS, data analysis and digitalization — designed for development practitioners in Africa and beyond.
Two articles are live — the rest is in preparation.
From realistic indicators to collection tools: how to build a monitoring & evaluation system that teams actually use — and that survives field realities, from patchy connectivity to staff turnover.
Design effect, intraclass correlation and the number of clusters required: what nested data structure changes.
NDWI, MNDWI, AWEI: what each index actually separates, and how to set a defensible threshold.
Reading a pixel's whole series rather than two images: principle, parameters and near real-time monitoring.
Censored data, Kaplan-Meier curves and proportional hazards: tracking pace rather than a final state.
Weighting criteria through pairwise comparison, checking consistency and testing the sensitivity of the resulting map.
Isochrones and cost surfaces: what Euclidean distance conceals, and the parameters that determine validity.
The parallel trends assumption, pre-trend checks and standard errors: what makes an estimate defensible.
Testing spatial structure at all scales, building the envelope correctly and choosing a defensible null model.
A structured qualitative method for documenting unanticipated effects — and the rigour conditions that make it defensible.
Non-parametric testing, robust slope estimation and the pitfalls specific to irregular satellite series.
Model structure, residual validation and interval reading: what to check before displaying a projection on a dashboard.
The Getis-Ord statistic tests the significance of a spatial cluster. Neighbourhood matrix, multiple testing and reading pitfalls.
Stratifying on what structures variance improves precision without raising the budget. How to build strata and allocate effort.
Multiresolution segmentation, object attributes and random forests: producing clean boundaries and areas you can stand behind.
Trend, seasonality, remainder: how to parameterise the decomposition and set alert thresholds that do not react to the calendar.
NDVI, SAVI, MSAVI, OSAVI: what each index corrects, and why the choice commits the comparability of a monitoring indicator.
Matching reduces selection bias on measured variables. What it corrects, what it does not, and how to check.
NBR, dNBR, RdNBR and RBR: what each index actually measures, and the caveats to observe before drawing a priority map.
KDE converts a point cloud into a continuous surface. Bandwidth, the decisive parameter, must be justified.
Architecture of a scheduled processing chain on Google Earth Engine, with quality controls and reliability conditions.
Equal intervals, quantiles, Jenks: how class breaks shape the reading of a choropleth map.
Using an eligibility threshold already in place to estimate a causal effect, without randomisation.
Moran's I, LISA and Getis-Ord Gi*: statistically testing whether an indicator's geography carries information.
Geostatistical interpolation based on the variogram, with a variance map showing where estimates are reliable.
Physical principles, Sentinel-1 processing chain and documented limitations of flood detection using radar imagery.
Satellite imagery, QGIS and open data: an accessible method to map forest dynamics in your intervention areas.
Too many indicators, not enough meaning: the most common Power BI pitfalls and how to avoid them to truly serve decision-making.
Why most digitalization projects fail for lack of understanding the real process — and how to avoid it in three workshops.
Design an XLSForm, deploy mobile data collection, manage your data and automate exports: the complete journey for your field surveys.
From your first map to spatial analysis: master QGIS to map your activities, beneficiaries and intervention areas.
Connect your sources, model your indicators and build a clear, automated project dashboard you can share with your partners.
Step by step, the construction of a mobile questionnaire with constraints, skip logic and data validation.
Guided tour of a real dashboard: page structure, visual choices and how project teams read the indicators.
Lessons learned from digitalizing multi-program monitoring & evaluation systems: steps, pitfalls and success factors.
Try another keyword, or suggest this topic — it might join the program.
Suggest this topicEditorial roadmap
Short, practical formats: MEL methods, dashboard pitfalls, process mapping. Enough to start the conversation.
Screen-by-screen demonstrations: XLSForm, Power BI, QGIS. Short, concrete, reproducible in your own projects.
Structured module-based journeys with exercises and datasets — to build skills end to end.
Follow me on LinkedIn to be notified as soon as the first content is out — or write to me if a topic particularly interests you.