Field guide
Design a small pilot with a clear stopping point
A pilot should buy useful learning within an agreed limit. Write the question, resources and review decision before inviting participants, so a busy trial does not become an open-ended service by accident.
Choose one uncertainty to test
Consider a fictional mentoring service testing evening sessions for people who cannot attend during working hours. Its first question is whether it can deliver those sessions consistently with the mentors and rooms available. That is different from asking whether mentoring causes sustained employment. A small delivery pilot can inform the first question without answering the second. Name the question precisely and resist adding every ambition of the organisation to the same trial.
State who the pilot includes and who it leaves out. Describe how people will hear about it and what they will be told. If invitations go only to existing members, the findings describe those members. They do not establish demand among people who have never used the service. Make this limit visible before anyone turns the pilot into a promotional case study.
Set the delivery boundary
Write down the number of sessions, trial period, available staff time and spending limit. Include preparation, administration, accessibility support and follow-up. Volunteer time still has a capacity limit even when it creates no invoice. Identify a person who can pause recruitment if the team cannot deliver the promised sessions. Do not let a funding deadline become the only reason the pilot continues.
Give participants a clear explanation of what happens when the trial ends. Record existing commitments that must still be honoured if the team stops recruiting. A stopping point is a management decision, not a licence to abandon participants. Check the practical handover before the first session: who answers questions, where people can find alternatives and who closes the attendance record.
Use a small record that people can maintain
For the fictional mentoring pilot, keep planned sessions, delivered sessions, attendance, cancellations, staff time and participant feedback in separate fields. Record why a session was cancelled instead of adding it to the delivered count. Do not put sensitive personal details in a shared operational sheet. Collect each field for a defined review question and agree who can access it.
Add a short note for exceptions. A session moved online after a room closure is not identical to the planned in-person session. If delivery changes, record the date and what changed so the team can interpret feedback in context. These notes are often more useful than adding another rating scale. The record should describe the trial the team actually ran.
Decide whether to stop, change or extend
Before launch, agree the conditions that require a pause, such as unavailable supervision or an inability to keep participant commitments. At the review, compare delivery against the plan, examine feedback and account for missing information. There is no universal success threshold for every pilot. Choose criteria that match this service's decision and the resources at stake, then document them before examining results.
End with one explicit decision: stop; change the delivery model and test again; or extend within a new limit. An extension needs a fresh question and resource decision. Successful delivery alone does not prove a long-term social outcome or financial sustainability. Use the separate outcomes record when preparing a report, and keep the pilot's limits beside any account of what it achieved.