Three mandatory fields fell below the agreed 95% operational threshold.
DEMONSTRATION CASE USING SYNTHETIC DATA
Field signals become
reviewable actions.
A four-day Kobo/ODK monitoring demonstration for a fictional household survey. No client, participant, programme outcome or accusation is represented.
Which field-quality signals require clarification today, and what evidence is needed before the next round of collection?
Signal register
Prioritise without pretending certainty.
The monitor compares agreed rules with incoming records. It does not infer intent and it does not silently delete or correct a submission.
Interviews below the context-specific duration threshold entered review; none was auto-rejected.
Low-accuracy or repeated-coordinate signals required field context; GPS was not treated as proof of presence.
Concentrated category patterns prompted back-check sampling rather than an accusation.
Field response
Each signal receives an owner, evidence request and deadline.
Data manager
Confirm the deployed form version and whether the three completeness gaps follow intended skip logic.
Field supervisor
Review the 18 short interviews with call logs and team context; select a bounded back-check sample.
M&E lead
Classify material issues, approve any correction rule and document limitations for the indicator report.
Analyst
Freeze a versioned export only after open quality decisions are resolved or explicitly carried forward.
Evidence boundary
What this demonstration does not claim.
- Repeated coordinates do not prove that an interview did not happen.
- Short duration does not prove fabrication or poor interviewing.
- Enumerator differences can reflect assignment, population or device conditions.
- A static export cannot reconstruct network/IP evidence that was never collected lawfully at the time.
Start with one project, one form and agreed review rules.
RMA can scope the minimum secure connection, thresholds, reviewer roles, issue register and management summary before live monitoring begins.