What are Semi-Structured Interviews?

A semi-structured interview is a one-to-one conversation guided by a prepared set of topics and open-ended questions, in which the interviewer is free (expected, in fact) to vary the order, improvise follow-ups, and pursue unexpected threads. It sits between two purer forms. The structured interview asks identical questions in identical order to every participant, maximising comparability and killing discovery; it is essentially a spoken survey. The unstructured interview follows the conversation wherever it flows, maximising discovery and complicating comparison. The semi-structured format keeps enough consistency that answers can be compared across participants, and enough freedom that the most interesting thing said in the session doesn't die for lack of a follow-up. That balance is why it is the default form of the user interview.

The Discussion Guide

The instrument is the guide, and building one well is half the method. A working shape: a warm-up that gets the participant talking about familiar ground; three to five topic blocks, each opening with a broad invitation ("tell me about the last time you...") and listing optional probes beneath it; and a close that asks what you should have asked. The craft rules are consistent with everything known about response bias: questions about specific recent behaviour beat questions about habits in general, neutral phrasing beats anything that leaks the hoped-for answer, and one idea per question. The guide's most important property is restraint: it is a safety net and a checklist, not a script. If you read it aloud verbatim, you have run a structured interview with extra paper.

Moderating the Hybrid

1. Know the guide well enough to abandon it.
The freedom to deviate only pays if you can find your way back. Internalise the topic blocks so the conversation can loop naturally and nothing gets dropped.

2. Follow the energy.
When a participant leans in, slows down, or volunteers something off-guide, that is the interview announcing where its value is. Chase it with their own words ("you said it felt risky, what made it feel that way?") and return to the guide afterwards.

3. Keep probes neutral and small.
"Tell me more", "what happened next?", and silence do most of the work. Every richer probe risks planting content, which is the standing hazard of improvised questioning.

4. Track coverage lightly.
A glance-able checklist of must-cover topics protects comparability without dragging the conversation into administration. Skipping a block for a rich tangent is a judgment call; making it knowingly is the skill.

5. Pilot the guide.
One practice interview exposes the questions that confuse, the blocks that run long, and the openers that produce one-word answers. Guides are prototypes too.

Analysing the Output

The format's partial consistency shapes the analysis. Because every participant met the same core topics, cross-case comparison is legitimate; because every conversation took its own path, the transcripts are not parallel questionnaires and shouldn't be coded as if they were. Systematic thematic analysis across the set, with the guide's topics as a starting frame and room for themes the guide never anticipated, fits the data's actual structure. The improvised passages, the ones no guide predicted, are routinely where the best themes come from, which is the whole argument for the format.

The Benefits

Semi-structured interviewing combines discovery with comparability: surprises can be pursued, and answers can still be laid side by side. The guide makes quality less dependent on interviewer improvisation (and makes delegating interviews to teammates feasible), while the freedom keeps sessions human, which participants reward with candour. For teams, a good guide is also reusable infrastructure, improving with every study that refines it.

The Limitations

The format inherits interviewing's general limits: self-report filtered through memory and social desirability, small samples that suggest rather than measure, and heavy dependence on moderator skill, since the improvised half of every session is only as good as the person improvising. Partial standardisation also means partial comparability; participants who took different paths answered subtly different interviews, a fact honest analysis has to hold. And the guide can quietly ossify: a document that never changes across studies has stopped being a hypothesis and become furniture.

The Takeaway

The semi-structured interview earns its status as qualitative research's workhorse by refusing both purities: enough structure to compare, enough freedom to discover. Build a guide of open questions about specific recent behaviour, learn it well enough to leave it, probe with the participant's own words, and analyse across cases with a method that respects how the data was made. The best moments will be the ones the guide never saw coming, which is precisely why you brought one.