How to follow up after a pilot: questions, script, and evidence.

1. Match the timing to the behaviour, not the posting calendar.

Choose the follow-up point when you design the pilot. If the promised next step is scheduling one conversation, a short interval may be enough to ask whether it happened. If the intended change needs a longer period to become observable, a next-day message cannot answer the question. Write the rationale and actual dates rather than assuming one interval fits every project.

Separate immediate experience, later action, and sustained change. You can ask about clarity immediately after delivery, action at the first follow-up, and persistence in a later round if justified. If you change the timing or only reach participants who are easy to contact, record that. The missing part belongs in the report, not outside the denominator.

• Immediate question: could the person understand and use the experience?
• Later question: what did they actually do afterward?
• Persistence question: did the intended behaviour continue, if this was observed?
• Limit: what does this time window leave unanswered?

2. Make the invitation voluntary and separate from promotion.

Say who is asking, which experience you are discussing, what you want to learn, and how long the conversation should take. State that a negative answer is useful and that participation does not affect access to support. Offer a simple way to decline. Agree on follow-up during the original experience instead of surprising people with repeated messages.

Do not ask for a public testimonial in the same opening request. Research participation and publication permission are different decisions. A person who agrees to tell you what happened has not automatically agreed to appear on your website. Keep recording, attribution, and quotation permissions separate and use an appropriate process for your setting.

• Name the experience and why you are following up.
• Give a realistic time commitment and an easy decline.
• Explain how notes will be used and who can see them.
• Do not make care, belonging, or further help depend on a favourable response.

3. Ask for a sequence of events before asking for an opinion.

Begin with the participant's own account: “Since the session, what happened with the next step you chose?” Ask for a concrete recent example, then explore the sequence. If they say nothing happened, ask what got in the way without treating the answer as a failure they must defend. Keep the conversation focused on their experience rather than your hopes for the project.

The GOV.UK interview guidance supports open, neutral questions and concrete examples. Use the menu below as an original discussion guide, not a survey you must finish at speed. Follow the relevant detail and avoid supplying the explanation. “How much did the worksheet improve your confidence?” assumes an improvement; asking what changed, if anything, leaves the direction open.

• What did you do after the session or after opening the resource?
• Can you walk me through the last time you tried that next step?
• What, if anything, was different from how you handled this before?
• What got in the way, or made you decide not to continue?
• What else helped or influenced what happened?
• Did you use any part of the resource again? Which part, and when?
• Is there an example you can describe without sharing someone else's private information?
• What have I misunderstood, or not asked, about your experience?

4. Keep a response record without guessing what silence means.

Keep the original eligible group, invitations sent, people reached, responses received, and relevant usable observations on separate lines. A response can be valuable without answering every question. Record the date and why information is missing when known. Do not repeatedly contact someone until you get a favourable answer; respect their preference and the contact plan you agreed.

A non-response is not a positive result. It is also not automatically proof that no action occurred. Leave the outcome unknown. Do not reduce the denominator to respondents without saying so, and do not combine results from different follow-up windows as though everybody had the same chance to act. These distinctions keep the report useful to somebody who was not in the room.

• Eligible for this follow-up: [count and definition].
• Invited / reached / responded: [separate counts].
• Usable observation for this question: [count and source].
• Missing or partial information: [what is unknown and why, if known].
• Method and timing: [who asked, when, and whether prompting occurred].

5. Write the strongest narrow sentence, not the biggest one.

This example is hypothetical and does not describe SIX33 results. Six participants are eligible for a seven-day follow-up after a planning workshop. All six are invited, four respond, and two respondents report completing the next action they chose. One respondent reports no action and one gives an answer that does not establish completion. Two participants do not respond. The completion reports are self-reported, not independently verified.

An inspectable sentence is: “At seven-day follow-up, four of six invited participants responded; two reported completing their chosen next action. Two outcomes were unknown because of non-response, and one response did not establish completion. These reports were not independently verified.” Keep the no-action response in the underlying record as well. Do not rewrite this as “our workshop changed a third of participants' lives” or assume the workshop caused the actions.

Ask what else might explain the result: prior plans, another resource, a helpful friend, or the person's own effort. An account can support a reported sequence without isolating a cause. For higher-stakes impact claims, get evaluation expertise appropriate to the question; a brief follow-up is not a substitute for a design that can support the inference you want to make.

6. Close the loop with the participant and the next version.

Thank the person for the actual detail they shared, not for confirming the mission. Explain what you will do next and honour the boundaries you stated. Keep a direct quote distinct from a paraphrase. If a story could identify somebody indirectly, removing a name may not make it safe to publish; choose a less identifying summary or leave it out.

Translate the findings into one change you can inspect. For example: participants understood the first question but did not know how to choose a date, so the next version adds a concrete scheduling example. Keep the old and new versions identifiable. A useful follow-up ends with a clearer product or a better question, not merely another paragraph of praise.