How Long Is a Research Interview? Length Benchmarks by Type

There is no single number, because duration tracks the interview type rather than some universal average. The most common qualitative format, the semi-structured in-depth interview, runs 30 minutes to several hours per DiCicco-Bloom and Crabtree (Medical Education, 2006). Survey interviews sit at the short end; oral histories run longest over repeat sessions.

So the honest answer is a set of benchmarks, ordered shortest to longest by type:

- Structured / survey interview: about 20 minutes (Pew Research Center, 2017) - Elite / expert interview: 20–45 minutes (Duke Initiative on Survey Methodology) - Oral history: 40–60 minutes per session (Guilford College) - Focus group: 60–90 minutes (Leung & Savithiri, 2009) - Semi-structured / in-depth interview: 30 minutes to several hours (DiCicco-Bloom & Crabtree, 2006)

Pick the wrong benchmark and the project math breaks. A team that assumes 20 minutes per session but actually runs in-depth interviews can under-budget its recording and transcription time threefold. Match the number to the method before you plan anything else.

Structured survey interviews are the shortest type, at about 20 minutes for a standard national telephone poll, per Pew Research Center (2017). Every respondent hears the same questions in the same order, so the conversation has little room to run long. Fixed script, fixed length.

Length here is a design decision, not a discovery. Structured interviews exist to compare answers across hundreds or thousands of people, so teams cap the questionnaire to protect response rates. Pew reports the 20-minute figure alongside no advance mailing, a detail that shows how carefully survey teams manage the burden they place on respondents.

Treat 20 minutes as the practical floor. Anything shorter is usually a screener or a street intercept, not a full research interview. If you ever see a much lower average-recording number, check whether short surveys are dragging the mean down; the mix of formats explains most of the gap, as average recording length breaks down.

Elite and expert interviews run 20–45 minutes, because access is scarce and asking for more time lowers your response rate, per the Duke Initiative on Survey Methodology. A 2026 methods guide built on over 100 health-policy interviews (Zavattaro and Gille) settled on about 30 minutes as its working norm.

Why so short, when the questions are open-ended? Because these are usually semi-structured interviews, not fixed surveys, and the mode shifts the clock. Bogner, Littig and Menz report that telephone interviews with experts tend to run shorter than face-to-face ones, so a phone slot with a busy official compresses the conversation further.

The limit is the calendar, not the material. You could talk with a senator or a chief executive for two hours; you rarely get the slot. Plan elite studies as many short, dense sessions, and bring a tight guide so 30 minutes still returns usable answers. Preparation buys back the time access takes away.

Semi-structured in-depth interviews, the workhorse of qualitative research, run 30 minutes to several hours, per DiCicco-Bloom and Crabtree (Medical Education, 2006). Oral history belongs to this family but stretches longest, because researchers collect it as a series of sessions rather than a single sitting.

For oral history specifically, three university library guides converge on a per-session window: - Ideal length 40–60 minutes; narrators tire by 1.5 hours (Guilford College) - No single session longer than 2 hours (Binghamton University) - Sessions capped at 60 minutes, never more than one per week (James Madison University)

Add those sessions up and oral history becomes the heaviest recording load per narrator. Someone interviewed weekly across several 60-minute sittings can generate more audio than a batch of one-hour in-depth interviews. Because the files pile up fast, plan the text step before you record; the per-file workflow is covered in how to transcribe an interview.

Focus groups run 60–90 minutes and seat 7 to 10 participants, per Leung and Savithiri in Canadian Family Physician (2009). The session runs longer than a one-to-one interview because the moderator has to give every person room to speak and to respond to each other.

Group size and study size both drive the total. Krueger's standard methods text puts a group at 5 to 8 people, ranging from 4 to 12, with at least three groups needed per study. One group tells you almost nothing; the pattern only shows up across several.

That pairing is the hidden cost. Three 90-minute groups is 4.5 hours of audio, but with eight voices overlapping it behaves like far more work to clean up than the raw time suggests. If focus groups are your method, plan capture and cleanup together; how to transcribe a focus group covers the multi-speaker case.

Every recorded hour costs far more than an hour to turn into text. Verbatim transcription takes 3 to 10 hours of work per hour of audio, per Bailey in Family Practice (2008). A common .edu budgeting rule sits inside that range at 4 hours per hour of audio.

Length also converts into word count, which sets editing effort. People speak at about 150 words per minute in conversation, per Baruch College. So one 60-minute interview is roughly 9,000 words at that single-speaker ceiling; real two-speaker dialogue with pauses nets closer to 7,000 to 8,000. Treat that word count as a derived estimate, not a published figure.

Scale it to a project and the load gets real. As an illustrative case, 20 in-depth interviews at one hour each is 20 hours of audio, which at the 4:1 rule is about 80 hours of manual transcription. That total is derived arithmetic from two sourced inputs, not a single published stat, but it shows why teams plan recording and cleanup together.

This is where interview length, study size, and transcription budget meet. For the totals across a whole project, see how many hours of audio in a qualitative study. Automatic speech-to-text shrinks the 4:1 ratio sharply, but you still review the output, so length never stops mattering.