Semi-Structured Interviews: The Complete Guide

Semi-structured interviews are the gold standard for qualitative research: structured enough to stay focused, flexible enough to follow where the conversation leads. With AI platforms like Koji, the semi-structured interview method scales to hundreds of participants without losing the depth that makes it so powerful.

What Is a Semi-Structured Interview?

A semi-structured interview sits between two extremes. A fully structured interview follows a fixed script — every participant gets the exact same questions in the exact same order. A fully unstructured interview is more like an open conversation with no predetermined agenda. A semi-structured interview gives you the best of both: a core set of questions you plan to cover, combined with the freedom to probe deeper, follow tangents, and ask spontaneous follow-up questions based on what participants say.

This flexibility is precisely why semi-structured interviews are the most widely used format in qualitative research. According to a 2023 survey by the User Research Institute, 67% of UX researchers rely on semi-structured interviews as their primary data collection method. They're particularly valuable when you're exploring a topic where you have some hypotheses but want to leave room for unexpected discoveries.

Think of it like this: a structured interview is like a multiple-choice test — you get consistent data but miss anything not on the form. A semi-structured interview is more like a guided conversation — you arrive prepared, but you're ready to be surprised.

How Semi-Structured Interviews Work

The core structure of a semi-structured interview involves three layers:

1. An interview guide — a set of 5–10 core questions you plan to ask every participant. These aren't scripts; they're anchors that keep the conversation on topic.

2. Probe questions — follow-up questions triggered by participant responses. These might be "Can you tell me more about that?" or "What do you mean by X?" They're not scripted in advance but flow naturally from what the participant says.

3. Emergent questions — questions you didn't plan but arise mid-conversation because a participant introduces a genuinely interesting angle. Good researchers note these and may add them to future interviews if they keep appearing.

As sociologist Dr. Steinar Kvale, author of InterViews: An Introduction to Qualitative Research Interviewing, wrote: "The purpose of the qualitative research interview is to understand the world from the subjects' points of view, to unfold the meaning of peoples' experiences."

This is why semi-structured interviews are so effective for product research, UX work, customer discovery, and academic inquiry alike.

How to Design a Semi-Structured Interview Guide

A great semi-structured interview guide is deceptively simple. Here's how to build one:

Step 1: Define Your Research Question

Before writing a single question, articulate exactly what you're trying to learn — not "I want to understand our users," but something specific like "I want to understand why enterprise buyers hesitate at the procurement stage."

Your research question guides every choice: who you talk to, what you ask, and what themes you listen for.

Step 2: Write 5–10 Core Questions

Core questions are the anchors of your guide. They should:

Be open-ended (start with "How," "What," "Tell me about," or "Walk me through")
Cover the key themes from your research question
Flow logically from broad to specific
Avoid leading language ("Don't you think...?" or "Wouldn't it be better if...?")

Example core questions for a product discovery study:

"Walk me through your current workflow for [task]."
"What's the most frustrating part of this process for you?"
"Tell me about the last time you tried to solve this problem."
"How do you decide which tools to use?"
"What does success look like in your role?"

Step 3: Prepare Probe Clusters

For each core question, brainstorm 2–3 probes you might use depending on what the participant says:

"Can you give me a specific example of that?"
"What happened next?"
"How did that make you feel?"
"What did you do about it?"
"Who else was involved?"

These aren't scripted — they're a mental toolkit you draw from when the conversation warrants.

Step 4: Include Warm-Up and Wrap-Up

Always start with 2–3 easy, non-threatening questions to build rapport. End with an open invitation: "Is there anything we didn't cover that you think would be helpful for me to know?"

Step 5: Time Your Guide

A typical semi-structured interview runs 30–60 minutes. With 5–8 core questions, plan 5–8 minutes per question, factoring in natural probing. Pad your estimate — participants often have more to say than expected.

Running a Semi-Structured Interview

The interview itself is where skill matters most. Here's what separates good semi-structured interviewers from great ones:

Follow the participant, not the guide. If a participant says something unexpected and interesting, pursue it — even if it's not in your guide. That unexpected thread might be your most important finding.

Use silence strategically. After a participant finishes speaking, count silently to three before jumping in. People often have more to say; they just need a moment. Silence communicates patience and draws out richer responses.

Probe on specifics, not generalities. When a participant says "I usually do X," ask "Can you walk me through the last time you did that?" Real incidents reveal far more than abstract descriptions.

Avoid "why" questions when possible. "Why did you do that?" can feel interrogative. "What led you to that decision?" covers the same ground with more warmth.

Take sparse notes. Your job is to listen and probe, not transcribe. Capture a few keywords to jog your memory later. If you're recording, trust the recording for specifics.

Analyzing Semi-Structured Interview Data

Because semi-structured interviews produce unstructured qualitative data — transcripts, notes, recordings — analysis requires a systematic approach.

Thematic analysis is the most common method: identifying recurring patterns, themes, and insights across multiple interviews. You read through transcripts, code relevant passages, and group codes into themes.

Manually, this process can take hours per interview. For a study with 15 participants, expect 40–60 hours of analysis work. That's why modern research teams increasingly use AI-assisted analysis tools. Platforms like Koji automatically extract themes, sentiment, and key quotes across all your interviews, compressing weeks of manual work into minutes.

Research from Nielsen Norman Group found that "automated synthesis tools reduce analysis time by 60–80% while maintaining comparable insight quality to manual methods" when used with well-conducted interviews.

When to Use Semi-Structured Interviews

Semi-structured interviews are the right choice when:

You're exploring a new problem space without strong hypotheses
You want depth and nuance, not just numerical data
You're validating a product concept or feature direction
You need to understand the "why" behind user behavior, not just the "what"
You're doing customer discovery for a new product or market

They're less appropriate when you need statistically representative data from a large sample (use surveys), or when you're testing specific UI/UX (use usability testing).

How AI Is Transforming Semi-Structured Interviews

Traditionally, semi-structured interviews required a trained moderator — someone skilled at probing, building rapport, and navigating unexpected directions. That meant high cost, scheduling complexity, and throughput limitations.

AI-powered platforms like Koji change the equation entirely. Koji's AI interviewer conducts semi-structured interviews autonomously: it follows your interview guide, asks your core questions, and generates contextual follow-up probes based on what each participant says — all in real time. Participants can interview by voice or text, on their own schedule, with no moderator present.

This means you can run 50 semi-structured interviews in the time it would take to schedule and conduct five manually. The AI captures full transcripts, analyzes themes automatically, and surfaces insights with citations back to specific participant quotes.

For teams doing continuous discovery, market research at scale, or studies across multiple time zones, AI-moderated semi-structured interviews remove the biggest bottleneck: researcher time.

Key Things to Know

Sample size: 5–15 participants typically reach thematic saturation for most qualitative studies. You're looking for depth, not breadth.
Recording consent: Always obtain explicit consent before recording. In semi-structured interviews, transcripts are essential for analysis.
Moderator effect: Your presence shapes the interview. Train yourself to stay neutral — avoid nodding excessively, reacting emotionally, or signaling preferred answers.
Triangulation: Semi-structured interviews are stronger when combined with other methods — surveys for scale, observation for context, or document analysis for corroboration.

Tips & Best Practices

Run a pilot interview before your main study to test your guide
Use a consistent warm-up sequence to establish rapport
Share your guide with stakeholders before fielding — it aligns everyone on research goals
Debrief immediately after each interview while impressions are fresh
Keep your guide to one page — complexity breeds rigidity

Frequently Asked Questions

Q: How is a semi-structured interview different from an in-depth interview? A: "In-depth interview" is often used interchangeably with "semi-structured interview." If there is a distinction, in-depth interviews tend to be longer (60–90 minutes) and more exploratory. Both use open-ended questions and probe-based follow-up.

Q: How many questions should a semi-structured interview guide have? A: Five to ten core questions is the sweet spot for a 30–60 minute interview. Fewer, and you risk running out of content; more, and you rush through without probing deeply enough on any single question.

Q: Can I use semi-structured interviews in quantitative research? A: Semi-structured interviews generate qualitative data by nature. However, you can quantify aspects of the analysis — like counting how often a theme appears — to complement quantitative findings.

Q: What is the best software for semi-structured interviews? A: Options range from basic video calls to purpose-built AI research platforms. Tools like Koji offer automatic transcription, theme analysis, and report generation that reduce post-interview analysis time by up to 80%, and the AI handles probing dynamically so you don't need a trained moderator for every session.

Q: How do I ensure reliability in semi-structured interviews? A: Use the same core guide across all participants, have two researchers independently code a sample of transcripts and compare themes, and document your analytical process for transparency.

Product & Research

Revenue & Growth

Advisory & Services