Write interview findings that readers can check
Write interview findings readers can trace to the source. Includes a fictional example, a blank template, and review and consent checks.
The interviews are over. Your notes are full of useful detail, but a reader cannot check a conclusion such as “People struggle with handover” unless you show what was said, by whom, and where to find it. A findings document should let someone follow each claim back to its source without exposing more of a participant’s words than you have permission to share.
This is not a guide to every method of interview analysis. It is a way to write up the work after you have reviewed the conversations and identified possible findings. The test is simple: could a colleague distinguish an observation from your interpretation, see an exception, and find the source?
Start with the decision the document must support
Write the question at the top: “What should we change about handover between shifts?” is more useful than “What did the interviews say?” Note who will use the document and what they can decide. A manager choosing a trial needs a different level of detail from a researcher reviewing the strength of a claim. Keep the question visible while you draft; it helps you leave out vivid stories that do not answer it.
Record the scope before the findings: who was invited, how many people took part, when the interviews happened, what roles were included, and what you did not cover. If participants were selected because they had reported a problem, say so. Do not present their accounts as a measure of how common the problem is in the whole organization. An interview can show how a situation works for one person; it cannot, on its own, establish a population-wide rate.
Make a source trail before writing claims
Give each interview a neutral label such as P01 or P02. Keep a private index that connects the label to the person only if you need one, and limit access to it. In a working sheet, put one relevant passage per row. Record the label, a recording time or transcript line, a short paraphrase, the exact wording you may quote, and your note about what it might mean. Check the recording where a name, number, negation, or speaker label matters: transcripts can get these wrong.
Separate three things in that sheet: what someone said, what you think it suggests, and what you propose to do. “P02 said they look for the latest update in two places” is an account. “The update location is unclear” is an interpretation. “Use one handover page” is a proposal that needs its own test. Keeping those apart stops a recommendation from masquerading as a fact.
For more on preparing accurate source text, see the interview transcription guide. The transcript is a record to check, not proof that every conclusion drawn from it is sound.
Write a finding that can be checked
A useful finding contains a bounded claim, source references, a short explanation of why those sources support it, and any evidence that points the other way. Say “In these interviews, several participants described checking more than one place for shift updates,” rather than “Everyone misses updates.” If you did not count accounts in a consistent way, avoid “most” or “80%.” If you did count, give both the number and the denominator, and explain which interviews you counted.
Use direct quotes sparingly. A quote illustrates an account; it does not replace an explanation. Include a time or line reference so a reviewer with access can read the surrounding exchange. A paraphrase may be safer than a quote when wording could reveal the speaker. Mark your paraphrase as a paraphrase, not as a quotation.
A sample findings document
Illustrative example only: the participants, passages, and situation below are fictional. They are not research results or customer testimonials. Imagine four interviews about shift handover at an invented service desk. The references show what a working document might contain; there are no real recordings behind them.
- Question and scope: How do staff find the latest handover note? Four fictional staff members, P01–P04, interviewed after a change to the shift rota. This small, selected group cannot tell us how often this happens across all shifts.
- Finding 1 — locating the current note takes extra steps for some staff: P01 (fictional transcript, lines 42–49) described opening both a shared folder and a message thread before a shift. P03 (lines 88–94) described asking a colleague which version was current. These accounts support a concern about where to look, not a claim that notes are missing.
- Disagreement and limit: P02 (lines 31–36) said the folder alone was enough for their shift. P04 did not discuss where notes were stored. We cannot infer that P04 had the same problem. The difference may relate to shift or access, but these interviews do not establish why.
- Possible next step: Ask staff on other shifts where they find the current note; then try a single, clearly dated note location with a small group. Check whether they can find the current note without asking someone. Do not call the trial a proven solution.
A reviewer can challenge this example: Are lines 42–49 really about finding the current note? Did P02 use the same folder? Would the proposed change help outside the interviewed shifts? That is the point of retaining the source references and the exception.
Use this blank template
Copy these headings into a document. Fill them with your own sources; leave a heading blank rather than inventing support.
- Decision question: What choice will this document help someone make?
- Scope: Dates, participant selection, roles, number interviewed, and gaps.
- Finding [number] — bounded claim: What did these interviews show, and for whom?
- Supporting accounts: Participant labels; recording times or transcript lines; permitted short quotes or marked paraphrases.
- Different accounts: Who described something else? Who was not asked? What remains unclear?
- Interpretation: What may explain the accounts? What other explanation could also work?
- Action to consider: What is worth trying or checking, who owns it, and what would count as a useful result?
- Review and sharing: Who checked the sources, what consent limits apply, and who may see this version?
Review the draft without losing trust
Ask a colleague to check a sample of claims against the source material, including at least one disagreement. They should verify speaker labels, quotation wording, line or time references, and whether a passage was taken out of context. If a passage cannot be found, remove or revise the claim. If two people read it differently, record that uncertainty rather than forcing agreement. Distinguish “not mentioned” from “denied”: silence is not a contrary account.
Check your consent terms before you share anything. Permission to record an interview does not automatically permit a public quote, a named attribution, or reuse in another document. Remove names and identifying details when the audience does not need them. In a small team, even a role or unusual event can identify someone; consider a broader description or a paraphrase. Keep the private source index and raw recordings apart from the shared document, and follow your organization’s rules for access and retention. If someone requested review of their quote or restricted its use, honor that request before publication.
Finally, let a stakeholder read the short findings and proposed next steps, then keep the source references in a restricted version for reviewers who need to check them. The result is not a polished claim that asks for trust. It is a clear claim with a route to verify it, an honest account of where people differed, and a next step that can be tested.