Automatic Kobo sync
Submissions and audit logs are pulled in and merged on a schedule, so the dashboard reflects what the field sent today.
Know your survey data is sound — before the team leaves the field.
Field Compass pulls your KoboToolbox submissions and audit logs, runs your validation rules against every record, and shows you exactly which submissions, which questions, and which enumerators need a second look.
Open source and self-hostable — your data never leaves your infrastructure.
Connect, don't migrate
Point Field Compass at your existing KoboToolbox project. Nothing about your forms or workflow changes.
Rules in plain English
Describe a check the way you'd say it out loud. Field Compass turns it into a rule that runs on every submission.
Evidence, not hunches
Audit logs reveal interview duration and post-submission edits — the signals a spreadsheet export throws away.
Trusted by teams at the world’s leading humanitarian and development organisations
Works with any KoboToolbox project — the collection platform these teams already run in the field.
What you get
Everything you need to trust your data
One place to define quality, spot problems early, and act on them while corrections are still cheap.
Validation rule builder
Compose range checks, required fields, skip-logic and cross-question conditions without writing code — then reuse them across rounds.
AI that writes your rules
Write “flag any survey completed in under 10 minutes” and get a structured rule back. Or let Field Compass read your form and propose the checks itself.
Quality overview
Pass rates, issue trends over time, and validation status at a glance — and which questions attract the most flags, so a wording problem doesn't get mistaken for a field problem.
Enumerator performance
Compare interview duration, flag rates and output per enumerator, so retraining goes to the people who actually need it — and nobody gets a lecture for a badly worded question.
Edit & history tracking
See which answers changed after a submission was first sent, when, and by whom — the audit trail your funder will ask about.
How it works
Running in an afternoon
No data migration, no new collection app for your team to learn.
1
Connect KoboToolbox
Add your Kobo API token — stored encrypted, per user — and pick the form you want to watch. Field Compass reads submissions and audit logs from there.
2
Set your quality rules
Build checks in the rule editor, describe them in plain English, or start from the set the AI suggests after reading your form.
3
Monitor and act
Watch the quality dashboard as data lands and review what gets flagged. Follow up with your field teams on the issues that come up while the enumerator is still in the area — when a correction is a phone call, not a return visit.
The quality loop
What happens to every record that lands
A submission arrives, gets checked, and — if something looks wrong — reaches the person who can still fix it.
-
Submitted
#4821 arrives from the field, audit log attached — timings, edits and all.
-
Checked
Your rules run against it, and the open-ended answers go to the review queue.
-
Flagged
Completed in four minutes. It surfaces on the dashboard with the reason attached.
-
Followed up
You reach the enumerator while they're still in the area, and mark it resolved.
The whole loop closes in hours, not at the end of the round.
AI agents
Put AI to work on your data
Writing validation logic is the part that stalls a quality plan. Hand it over: Field Compass reads your questionnaire, drafts the rules, and keeps working through incoming answers while you get on with the survey.
It writes the rules. “Flag if respondent age is greater than 100” becomes a structured rule that runs on every submission from then on.
It reads your form first. Field Compass analyses your questionnaire and proposes the range, consistency and required-field checks it expects you'll need — before you've written a line of logic.
It keeps working in the background. Qualitative checks run on their own queue, reading open-ended responses for empty, off-topic or copy-pasted answers while you do something else.
It reports back before it acts. Every rule it drafts arrives as a plain condition you can edit, test or discard. It proposes; you approve.
You describe it
Flag any survey completed in under 10 minutes
{
"name": "Interview too short",
"field": "active_interview_time",
"operator": "less_than",
"value": 600,
"severity": "high"
}Start watching your data quality today
Create an account and connect your first Kobo form in minutes.