A researcher asks how a manager prepares a weekly rota, expecting a conversation about scheduling software. The manager starts talking about the messages staff send on Sunday evenings: a changed childcare arrangement, an exam, a request to swap shifts. Following that account could reveal something important about the job, but the researcher still needs to cover the questions that brought them to the interview.
Semi-structured interviewing accommodates this kind of departure. It uses a prepared set of topics while allowing the interviewer to change the order, ask follow-ups, and explore something unexpected. The format is useful when the study needs both a common basis for comparison and enough flexibility to understand experiences the team has not fully anticipated.
What stays consistent, and what can change
The structure comes from the research objectives and the topics each participant should encounter. In the fictional rota study, those might include gathering availability, making changes, communicating the schedule, and resolving disputes. The manager’s Sunday messages belong within that inquiry, even if the guide did not predict them.
Flexibility concerns how those topics are explored. A participant may cover communication while describing a recent change, making a later question unnecessary. Another may need more context before they can give a useful example. Record these differences rather than assuming that every person answered an identical set of questions simply because the sessions used the same interview guide.
A structured interview offers tighter standardisation, while an unstructured interview leaves more of the direction open. Neither is inherently shallow or superior; the choice depends on the question and the kind of comparison the research requires.
Following up without supplying the answer
A useful follow-up begins with something the participant has actually said. If the manager describes a change as “awkward”, ask what made it awkward or what happened next. Asking whether staff were angry introduces an explanation that may not have been part of the experience at all.
This becomes more demanding when the interviewer knows the product well. Familiarity can help them understand a reference, but it also makes it tempting to finish an explanation or suggest a workaround. Leave room for the participant to describe their understanding before clarifying the details that matter to the research. A pause can help them think, though it is not evidence that the answer which follows is more truthful.
The GOV.UK interview guidance describes the value of a flexible discussion guide. In practice, that flexibility works best when the interviewer knows which topics are essential and can return to them without abruptly interrupting the participant’s account.
Keeping the study coherent across sessions
A pilot helps identify questions that are difficult to understand and topics that need more time. During an exploratory study, later sessions may benefit from revised prompts as the team learns, but those changes need a record. An issue raised spontaneously in the first interview and explicitly prompted in the next six has not emerged in the same way across all seven.
The same care applies when several researchers share the interviewing. Discuss the purpose of each topic, the kinds of neutral probes that may help, and how to document deviations. Consistency is easier to maintain when colleagues understand what a question is trying to establish instead of merely copying its wording.
Analysing uneven conversations
Start by examining the accounts in their own context before comparing them. One participant may spend considerable time on an issue because it is important to their work; another may never have encountered it, or may not have been asked about it. Silence in a transcript cannot automatically be counted as absence of the problem.
A systematic analysis should preserve those distinctions while developing patterns across the material. The user-interview overview explains the wider evidential limits, and thematic analysis provides a route into working with the accounts. The value of the semi-structured format lies in the combination: the team can investigate a shared question while still learning why people’s experiences differ.