Ideas, lessons and tools from our work.

How to write an indicator reference sheet people will actually use

An indicator can look clear in a logframe and still be measured five different ways. A simple reference sheet gives everyone a common definition, method and interpretation.

A result indicator helps a team monitor change, but the indicator name alone rarely provides sufficient instruction. Take “percentage of households with acceptable food consumption.” Different field teams could use different recall periods, household definitions or thresholds and still believe they are measuring the same thing. The result is a dataset that looks consistent but is not genuinely comparable.

An indicator reference sheet makes these choices explicit before data collection begins. A practical sheet should identify the indicator title and code; definition and calculation; unit of measurement; numerator and denominator where relevant; source and collection method; population or sampling approach; disaggregation; reporting frequency; baseline, targets and responsible person; quality checks; and important interpretation notes.

Start with the management question

Ask what decision the indicator will support. A metric for operational monitoring may need frequent, low-cost collection. An outcome indicator might require a periodic survey with greater methodological care.

Specify the calculation

For percentages, write out the numerator and denominator. Explain which respondents qualify, how missing responses are treated and whether weights will be used.

Takeaway: A good indicator reference sheet is not an appendix to ignore. It is a practical agreement about what the result means and how the team will know.

Discuss your measurement priorities

Why evaluation reports go unread—and what to do about it

A rigorous evaluation can still have limited influence when its findings arrive too late, overwhelm readers or fail to connect with real decisions. Evidence use needs to be designed, not assumed.

Evaluation reports often serve several audiences at once: funders need accountability, implementers need operational learning and decision-makers need clear options. A single lengthy document rarely serves each audience equally well.

Begin with the decisions

During inception, ask which choices could reasonably change because of the evaluation. Will the organisation redesign activities, expand a model, change targeting or revise a budget?

Produce layers of communication

Maintain the complete technical report, but accompany it with a short decision brief, clear visual summaries and a presentation tailored to the intended audience.

Make recommendations actionable

Replace vague advice such as “strengthen coordination” with a concrete action: establish a coordination meeting, identify its owner, specify the cadence and define how performance will be reviewed.

Takeaway: The measure of a useful evaluation is not simply whether the final report was accepted. It is whether the evidence reached the right people in a form they could understand and use.

Plan an evaluation with evidence-use milestones

A practical daily checklist for survey data quality

Data quality cannot be fixed entirely at the analysis stage. Simple checks during fieldwork can catch avoidable errors while the team can still respond.

Every day of fieldwork creates decisions: which interview should be revisited, which enumerator needs coaching and whether the sampling plan is still on course. A short daily review makes those decisions visible before small issues accumulate into major data-quality problems.

Ten checks for field supervisors

1. Confirm completed interviews against the daily assignment and approved sample.
2. Review missing values in essential questions and investigate unusual patterns.
3. Check skip logic and consistency between related responses.
4. Identify implausible entries, extreme values and duplicate records for review.
5. Examine interview durations as flags for investigation, not automatic proof of misconduct.
6. Compare timestamps and device records with the approved fieldwork schedule.
7. Review geographic coverage and deviations from the sampling plan, respecting location privacy.
8. Check enumerator-level patterns for coaching needs or possible systematic errors.
9. Record corrections, callbacks, refusals and unresolved issues in a documented log.
10. Securely synchronise, back up and restrict access to data according to the approved plan.

Takeaway: Reliable datasets are built through consistent field practices, not rescued through last-minute cleaning.

Explore data and research services