Test an inspection form with missing, contradictory and repeated answers before asking field staff to rely on it.
A successful submission is not the same as a useful record
A hypothetical inspection form accepts an asset identifier, condition and follow-up note. The clean example works. The harder question is what happens when an inspector does not know the condition, chooses a different answer halfway through, or submits twice after an uncertain connection.
Survey123 supports form logic, including expressions and constraints. The following workbook defines expected business behavior; it is not executable XLSForm code or a claim that every check is built in automatically.
Write the expectation before testing
| Test input | Expected business result | What the reviewer inspects |
|---|---|---|
| Missing asset ID | No apparently complete inspection without its subject | Required behavior and message clarity |
| Value outside the agreed range | Clear explanation or approved exception path | Validation matches actual units and policy |
| “Unknown” condition | Explicit uncertainty, not “good” by default | Stored value and follow-up responsibility |
| Answer changed after a follow-up was entered | No contradictory retained answer | Resulting stored record, not just visible controls |
| Same inspection submitted again | Identifiable potential duplicate | Agreed identity rule and review process |
The duplicate row is especially important: a required field does not establish uniqueness across submitted records. Define what makes two observations the same inspection before selecting a technical control.
Work one failure all the way through
In this synthetic case, the inspector selects “defect found,” enters a follow-up note, then changes the condition to “unknown.” The reviewer checks whether the saved record still tells a coherent story. Different visibility mechanisms and app behavior can matter, so test the published form on each intended client rather than relying on the design preview.
The acceptance decision could be: keep the uncertainty, require an explanation appropriate to that state, and prevent an unrelated defect response from being misread. The form designer implements that agreed behavior and repeats the case; this article does not claim that a particular expression has been tested.
Give the Arabic version its own review
Ask a reviewer to interpret the Arabic prompt without seeing the English. Does “condition” mean physical condition or operating state? Are units visible? Can a technician distinguish “not inspected” from “no defect”? Use stable stored codes with clear localized labels where the design calls for them.
Keep a short result sheet with form version, client version, input, expected result, observed result and owner. A release is ready when material failures are resolved or explicitly bounded—not when every tester merely managed to press Submit.
For related preparation, read the offline field handoff. Discuss this workflow with InfoGraph or review relevant GIS training.



