Run a research readout with skeptical stakeholders

A research readout should help the team decide what the evidence changes. Skeptical questions are part of that job. “How do we know?” may expose a weak claim, a missing audience, or a disagreement about the decision rather than resistance to research itself.

Design the meeting around the decision and its uncertainties. A long sequence of impressive quotes can leave stakeholders entertained but no better equipped to choose what happens next.

GOV.UK’s sharing guidance recommends showing the research context, findings, consequences, and next steps so others can use and question the work. Read the sharing guide. The agenda below is an original facilitation tool for a small team.

Send the decision question before the meeting

A brief pre-read should state the choice, the alternatives still in scope, who will decide, and the evidence limits. Include the research question, participants’ relevant contexts, methods, dates, and missing coverage.

For a fictional inventory product, the decision might be: “Should we test clearer update history before committing to a full reconciliation dashboard?” That is narrower and more actionable than “What did customer research tell us?”

Ask attendees to bring a specific challenge or a competing explanation. Provide appropriately restricted source links for people who need to inspect the evidence. A pre-read should not expose raw participant material more widely than the study permits.

Use a 45-minute agenda with room for challenge

The schedule is an example to adapt, not a rule about how long every research meeting should take.

Time	Activity	Desired output
0–5 minutes	Restate the choice, owner, and limits	Agreement about the decision being considered
5–13 minutes	Present the few findings that affect that choice	Shared understanding of claims and source context
13–23 minutes	Examine counterevidence and alternative explanations	A visible list of substantive challenges
23–33 minutes	Sort what each challenge would require	Specific checks, corrections, or accepted uncertainty
33–41 minutes	Compare next actions and decide	Decision or a defined condition preventing it
41–45 minutes	Assign owners, dates, and revisit trigger	A record of who will do what next

If the evidence cannot support the proposed decision, changing the decision is a legitimate output. The purpose is not to secure approval for the researcher’s favored recommendation.

Present a finding with its boundary

Use one claim per slide or discussion card. Put the supporting observation, relevant context, counterexample, and implication together.

A fictional card might say: “In the studied multi-system inventory workflows, uncertainty about the latest update contributed to manual comparison.” It would show the source references, the contrasting case where update time was clear, and a separate case where comparison was required for approval.

The implication is to test the explanation and a possible remedy in that workflow. The claim does not establish how many organizations have the problem or whether a dashboard is the best solution.

GOV.UK’s analysis guidance separates observations, findings, and actions. Preserving that separation in the readout makes it easier to locate a disagreement. Read the analysis guide.

Turn objections into an evidence map

Write the challenge in neutral language and ask what observation would resolve or materially change it.

Challenge	What it may reveal	Useful response
“You only talked to a few people.”	The conclusion may be broader than the sample supports	Narrow the claim; identify whether the decision actually requires prevalence evidence
“Our largest customer does something different.”	A relevant context or segment may be missing	Compare their workflow and decide whether separate research is needed
“The dashboard says people complete this.”	Different definitions of completion or burden	Align task, date, unit, and event definition before claiming conflict
“This does not fit our strategy.”	A priority disagreement rather than a factual dispute	Ask the decision owner to state the tradeoff explicitly
“The proposed solution will not help.”	The recommendation goes beyond the finding	Separate the explanation from the intervention and design an appropriate test

Do not answer every challenge with more of the same evidence. A prevalence question may need a different method. A strategy dispute may need a decision from the person responsible for that tradeoff.

Keep the discussion from becoming a vote on anecdotes

Give attendees a brief opportunity to write their challenges before discussion. Invite different perspectives and ask people to point to the claim they are challenging. This is a facilitation suggestion, not a guarantee against group influence.

When someone offers a contrary case, record its context and source. It should be examined, not dismissed because it conflicts with the main pattern. Likewise, a forceful anecdote should not erase other evidence without inspection.

Use a visible status for each challenge: corrected now, needs a defined check, outside this decision, or unresolved and accepted by the decision owner. Avoid calling an issue “resolved” merely because the meeting moved on.

Close with an accountable decision

Record the chosen action, the reasoning, important dissent, owner, due date, and what result would trigger reconsideration. If no decision is possible, name the missing evidence and the person responsible for obtaining it.

For the fictional inventory example, the output could be a scoped update-history test and a separate check on approval-driven comparison. The team has not committed to a broad dashboard or claimed the research proved its value.

A good readout leaves stakeholders able to explain both the decision and its limits. The research remains useful when a later result changes the team’s mind, because the reasoning and the revisit condition were recorded.