A poor survey is expensive twice: it consumes the rep's time in the store and produces data that nobody trusts or uses. Strong field surveys start with a decision and work backwards to the few observations required to make it.
Start with the decision, not the question bank
Ask who will use the result, how often they will review it and what action it should trigger. If an answer does not change an operational decision, consider removing that question or moving it to a less frequent research exercise.
Create a question specification for each item:
| Design element | Example |
|---|---|
| Business decision | Should this store receive a restock intervention? |
| Question | Is the authorized priority SKU available on the intended shelf? |
| Response | Yes / No / Not applicable / Could not verify |
| Applicability | Only stores authorized for that SKU |
| Evidence | Relevant shelf photograph where permitted |
| Follow-up on failure | Quantity, context and likely root cause |
| Owner | Field manager or replenishment team |
Use consistent response semantics
Yes should always mean the statement is true. Avoid alternating between "Is stock missing?" and "Is stock present?" without clearly handling the difference in scoring. Provide not applicable for genuinely ineligible checks and could not verify for incomplete observation. These are not the same.
Use conditional follow-up questions so reps provide a cause or photo only when a problem exists. If a question can be answered from reliable master data, do not repeatedly ask the field team to type it in.
Separate measurement types
- Observation: present, absent, quantity, price, expiry date.
- Evaluation: compliant or not against a documented rule.
- Explanation: reason for an exception, preferably from a controlled list plus notes.
- Action: corrected, escalated, deferred, authorized exception.
Do not collapse these into one open-text field. A manager should be able to compare patterns across hundreds of stores without interpreting hundreds of writing styles.
Pilot in real stores
Observe a small set of representatives completing the survey in different store formats and connectivity conditions. Check completion time, ambiguous questions, skipped items and disagreement between observers. Ask whether the evidence genuinely supports the proposed score.
Look for double-counting: if the same question appears in several tasks, a single visit may produce seemingly contradictory answers. Decide whether it should be repeated, shared or reported separately with survey context.
Keep versions and instructions
When the scoring rule or question wording changes, retain a version history so a trend shift is not mistaken for improved performance. Communicate changes through simple field instructions and check comprehension, not only a release note.
Where FieldKo fits
Surveys in FieldKo are built in Salesforce, attached to tasks and answered in the app, with follow-up questions where you need them. Design the survey on paper first, then build it.
See how FieldKo supports this →Further reading
Next: Improve photographic evidence | Define your KPIs
Reviewed October 26, 2026. Examples, scores and thresholds are illustrative, not benchmarks. See the research and sources library for independent references.