Agriculture & Future Food Systems · Article
Pilot, Monitor and Obtain Feedback
Use a limited food-service change to learn about delivery, people's experience and practical burdens before extending it.
Preparation and related reading
Shared Foundations
Related articles
- Set a useful food-use goal and baseline
- Practise a food-flow record
- Compare food-use results and investigate causes
- Change one food-use step and review again
- Questions, Indicators and Causal Claims
- Household, Business and National Scales
- Survey, Administrative and Flow Data
- Trade-offs and Transparent Dashboards
This article helps you
- Specify what a small pilot can investigate and what it cannot establish.
- Track planned work, actual operation and record coverage separately.
- Use participant feedback without treating respondents as everyone affected.
Give the pilot a bounded purpose
A vegetable-box service wants to test evening collection. A pilot is a limited trial used to learn about an arrangement before a wider commitment. The useful question might be whether the intended users can collect during those hours and whether the service can staff them reliably.
Describe the existing baseline, the proposed change, the included people and the period. A short trial can reveal operating problems and plausible benefits without establishing long-term outcomes or a causal effect. Keeping the question small helps choose records that matter rather than collecting everything that happens.
Specify the record and review point
Decide how to record scheduled sessions, actual openings, receipts, cancellations and missing data. Keep household identities separate from repeated visits. Record relevant conditions and deviations so later readers can understand what was actually tested. Agree who collects and checks the data, and when the team will review it.
Set revision or stopping conditions appropriate to the service. A pilot does not remove food-safety, staff-competence or product-suitability requirements. Maintain an appropriate fallback for users if the arrangement cannot continue. Operational controls belong in their specialised topics; the plan should identify the evidence and responsibility they require.
Successful opening and complete measurement differ
In an invented pilot, eight sessions are planned. Seven open and one is cancelled. Receipt counts are available for six of the seven open sessions; one open session has a missing count. Opening coverage is 7/8=87.5%. Receipt-record coverage among open sessions is 6/7, approximately 85.7%.
These are implementation and data-coverage measures, not percentages of households receiving food. A cancelled session and an open session with missing receipt data are different events. Neither should be silently removed to make the record look complete. The available counts describe only the observed sessions unless a justified further method is used.
| Record | Count | Meaning |
|---|---|---|
| Planned sessions | 8 | Intended operation |
| Opened sessions | 7 | One cancellation |
| Open sessions with receipt counts | 6 | One open session lacks a count |
| Opening coverage | 7/8 =87.5% | Operational record |
| Count coverage among open sessions | 6/7 ≈85.7% | Data coverage |
Follow sessions and their records separately
Read the diagram as text
The invented pilot plans eight sessions, opens seven and retains receipt counts for six of those opened sessions. A cancellation differs from a missing count.
Parts
- Planned: 8 sessions
- Defined opening plan for the pilot period
- Opened: 7 sessions
- One cancelled; 7/8=87.5% opening coverage
- Counts: 6 of 7 open sessions
- One open session has an unknown receipt count
Connections
- Planned: 8 sessions → Opened: 7 sessions: check operation
- Opened: 7 sessions → Counts: 6 of 7 open sessions: check records
Feedback can reveal a missing mechanism
Ask users and workers what changed and what remained difficult. A customer might prefer the evening time but still lack transport. A worker might find staffing unpredictable. These accounts can suggest why an output did not produce the intended outcome, and can identify unexpected burdens requiring investigation.
Explain the purpose of collecting information and handle participation and personal data appropriately. Seek perspectives from nonusers or hard-to-reach groups as well as convenient attendees. Distinguish an individual's experience, a repeated pattern and a representative population estimate. Qualitative detail is useful evidence without needing an invented percentage attached to it.
Respondents are not the complete group
For a separate invented feedback record, 40 eligible households are invited and 25 respond. Five respondents say collection timing became easier. That is 5/25=20% of respondents, and response coverage is 25/40=62.5%. It does not establish that 20% of all eligible households benefited, that the other respondents had worse access, or that overall food access improved.
Retain the actual question and what the other answers meant. Lack of response is unknown information, not dissatisfaction or success. A report can combine the observed experiences with receipt and staffing records while preserving their different units and evidence limits.
Use the pilot to decide the next investigation
Compare actual operation with the plan and describe both useful changes and unresolved problems. Check whether the intended people participated and whether the conditions match a possible wider service. A trial at one accessible site during one season may not transfer to distant sites or a different staff team.
The next step can be to continue, revise or stop the trial. Explain which finding supports that choice and what further evidence is needed. Shared system and uncertainty principles support the interpretation; neighbouring evaluation and resource topics offer deeper methods. A pilot is a way to learn, not a guarantee of success at larger scale.
Worked examples
Read two coverage measures
What do the invented eight planned, seven opened and six counted sessions show?
- Calculate 7/8=87.5% opening coverage.
- Calculate 6/7≈85.7% receipt-record coverage among open sessions.
- Keep the cancellation and the missing open-session count as separate records.
The pilot operated seven of eight planned sessions, but receipt information covers only six of those seven. Neither ratio measures household collection success.
Describe feedback respondents
How should five positive timing responses among 25 respondents from 40 invited households be reported?
- Compute 5/25=20% among respondents.
- Compute 25/40=62.5% response coverage.
- Retain the question, other answers and the unknown nonrespondents.
Five respondents reported easier timing. This is 20% of respondents, not a measured benefit rate for all 40 eligible households.
Check your understanding
Should an open session with an unknown receipt count be recorded as zero receipts?
Show answer and reasoning
No.
The receipt outcome is missing, while the session's opening status is known.
Does one successful site prove the same arrangement will work in every season and location?
Show answer and reasoning
No.
People, staff, transport and conditions can differ; generalisation requires evidence.
Can feedback identify a barrier even when it is not a representative prevalence estimate?
Show answer and reasoning
Yes.
A specific experience can guide a useful follow-up while its scope remains explicit.
Key ideas
- Give a pilot one useful question, boundary and review point.
- Record implementation and data coverage separately from outcomes.
- Retain user and worker experiences with their actual coverage.
- Review unexpected burdens and transfer limits before extending the arrangement.
References
Sources supporting the explanations and examples in this article.