This manual is for the person responsible for a data collection operation: putting the right work in front of enumerators, watching quality as data arrives, verifying it, and closing the loop on what needs fixing. It assumes surveys are already built (see the survey building guide) and the team exists (see managing enumerators).
Putting work in front of enumerators
Field Operations → Assignments and Board manage collection targets: assign an enumerator to a survey with a target count, watch progress on the board, and rebalance when someone falls behind. Enumerators see their work in the mobile app's My Assignments tab.
Cases are for named-respondent work: when specific households, patients, or businesses must each be interviewed, enrol them as cases in bulk (from your respondent directory) and assign the caseload. Each case tracks its visit attempts against an attempt budget:
- Recording an attempt with refused, moved away, or ineligible closes the case immediately with that outcome.
- Unavailable, callback, and other count toward the maximum attempts; a case that runs out of attempts lands in the Exhausted tab for your decision.
- A completed interview closes its case automatically when the response arrives; nobody marks cases complete by hand.
- Reassigning a case to another enumerator preserves the full visit history.
Devices. For managed device fleets, the Devices page holds the inventory: each device's last check-in, app version, and pending (unsynced) response count, which is your early-warning signal for a device that has stopped syncing. You can mint and print QR enrolment codes from here today (including pre-bound codes that will sign an enumerator in with no typed credentials); the in-app scan step that consumes them arrives in an upcoming app release, so current setups still sign in with email invitations.
Watching data arrive
The Quality page is your daily view. It combines the review queue with automatic quality flags raised on incoming responses:
- Speeding: interviews faster than the thresholds you configure per survey (set the speed rules from the Quality page; they apply from then on).
- Straightlining: identical answers down a grid or scale battery.
- Duplicate identifiers and outliers on numeric answers.
- GPS variance: capture locations outside the expected area, when geofencing is configured.
Per-question forensics on each response (time spent, answer changes) help you distinguish a fast-but-honest enumerator from a fabricated interview.
The review pipeline
Each response moves through a review state, and only approved-quality data flows into analytics and exports:
- Approve a response, or reject one that is unusable.
- Correct an answer when the fix is known (a transposed number, a mis-keyed code). Corrections require a reason, keep the original value permanently in the revision history, and put the response back under review so the change itself is reviewable.
- Return to field sends a response back with a note when the enumerator must resolve it.
- Quarantine parks a response outside analytics while you investigate, without deciding yet. Quarantined and rejected responses are excluded from dashboards, indicator calculations, and default exports automatically.
If your systems need to know, review events (approved, rejected, returned, quarantined, corrected) can be delivered by webhook.
Back-checks
Verification is sampled, not ad hoc: from the Quality page, draw a back-check sample as a percentage per enumerator. The sample is risk-weighted, flagged and already-suspect interviews are drawn first, and the remainder at random, so your verification effort lands where it matters. Assigning a verifier turns each sampled interview into a verification case for the case workflow above. Record back-check results against the sample and the worklist shows what remains.
Closing the loop
When collection surfaces issues that need action (a complaint, a service failure, a follow-up promise), routing rules can open follow-up cases automatically from responses, with SLA timers and escalation. That workflow lives under Follow-up and is covered in the follow-up and CFM guide.
A day in the life, condensed
- Morning: Devices page for silent devices; Board for progress against targets.
- Midday: Quality page; act on new flags, correct or return what needs it.
- Ongoing: keep the Exhausted tab at zero by deciding those cases daily.
- Weekly: draw the back-check sample, review forensics on anything flagged, and check retention of quality across enumerators before it becomes a pattern.