Indicators and learning

An indicator is an observable signal about an outcome. Define the construct, unit or qualitative rubric, population, data source, collection cadence, owner and interpretation limits. Add a baseline and target only when supplied or deliberately agreed; “unknown” is better than an invented number.

Use disaggregation when aggregate results could hide unequal access, benefit or harm. Balance reach with quality, behavior with condition, leading with lagging signals, and intended with adverse outcomes. Consider burden, privacy, accessibility and whether data collection could itself exclude or harm people.

Turn the most consequential uncertainties into learning questions. Define what evidence will be reviewed, by whom, when, and which adaptation decision it could trigger. Monitoring an indicator does not establish attribution; evaluation design and causal inference remain separate professional tasks.