Semi-Structured Interviews: Configuration and Best Practices

A Semi-Structured Interview on Qualz.ai balances the consistency of a script with the flexibility of natural conversation. The AI moderator follows your core questions but can rephrase for clarity, adjust the order slightly, and ask a single follow-up probe per question when it adds depth. This is the most versatile interview mode and is suitable for the majority of qualitative research projects.

What Makes a Semi-Structured Interview Different?

The semi-structured mode gives the AI moderator controlled flexibility:

Questions can be rephrased — The AI may rephrase a question slightly to match the flow of conversation or to improve clarity, while preserving the original intent
Order can be adjusted — If the participant's response naturally leads into a later topic, the AI can adjust the sequence slightly for a more natural flow
One follow-up probe per question — After the participant responds to a main question, the AI can ask a single probing follow-up to explore the response more deeply
Smart skipping — If a participant's answer to one question already addresses a later probe or question, the AI will skip it to avoid redundancy

This makes semi-structured interviews ideal for studies where you need both coverage of key topics and the depth that comes from natural conversation.

How to Set Up a Semi-Structured Interview

Step 1: Select Semi-Structured Interview Type

During the interview creation process, select Semi-Structured from the Interview Type dropdown.

Step 2: Generate or Write Your Questions and Probes

Option A: Generate with AI Click Generate with AI in the Interview Guidelines section. The platform generates main questions, each with an optional probing question underneath. The AI allocates approximately 2-3 minutes per question-and-probe pair:

For a 15-minute interview, you might receive:

5-7 main questions
Each with an optional probe for deeper exploration

Example of generated output:

Topic 1 - First Impressions
1. What were your initial thoughts when you first started using the product?
   Probe: Can you describe what stood out to you the most?

Topic 2 - Daily Usage
2. How do you typically use the product in your daily workflow?
   Probe: Could you walk me through a specific example from yesterday?

Option B: Write Your Own When writing your own semi-structured guidelines, include both main questions and optional probes:

Main questions should be open-ended and focused on a single topic
Probes should dig deeper into the response, not be standalone questions
Mark probes clearly (e.g., indent them or prefix with "Probe:")

Step 3: Review Question-Probe Pairs

Before saving, check that:

Each probe genuinely extends the main question rather than introducing a new topic
The total number of main questions fits your time limit (allow 2-3 minutes per pair)
Probes are optional depth-seekers, not essential questions

How the AI Moderator Runs a Semi-Structured Interview

Opening

The AI introduces itself, briefly explains that this will be an open conversation, and immediately asks the first question from the guidelines.

During the Interview

For each main question, the AI follows this adaptive pattern:

Asks the main question (may rephrase slightly for natural delivery)
Listens to the participant's full response
Decides whether to probe:
- If the response is brief or surface-level, the AI asks the probing follow-up
- If the response already addresses the probe's topic, the AI skips it
Gives a brief acknowledgment and transitions to the next question

Key behaviors:

Asks either a main question or a probe per turn, never both
Can adjust question order if the participant naturally brings up a later topic
Skips redundant probes when the participant has already provided that depth
Never introduces questions that aren't in the guidelines

Closing

After covering all questions (or when the participant chooses to end), the AI thanks them and concludes.

When to Use Semi-Structured Interviews

Ideal scenarios:

User experience research — Understand how people interact with products while maintaining coverage of key topics
Market research — Explore customer needs and preferences with room for unexpected insights
Academic qualitative research — Standard methodology for most qualitative studies
Employee or stakeholder interviews — Cover required topics while allowing natural conversation
Any study where you have specific questions but want the flexibility to explore interesting responses

Consider a different mode when:

You need exact question-to-question comparability across participants — try Structured
You're in early discovery and don't have specific questions yet — try Unstructured
You have a very specific research methodology in mind — try Specialized Modes

Best Practices

Write probes that deepen, not redirect — A good probe extends the main question: "Can you tell me more about that?" A bad probe introduces a new topic: "What about pricing?"
Don't over-probe — Include probes only where deeper exploration adds value. Not every main question needs a probe. The AI will use them selectively based on the participant's response.
Trust the AI's judgment on skipping — The AI is designed to skip probes when the participant has already provided sufficient depth. This keeps the interview feeling natural rather than repetitive.
Use the Additional Instructions field strategically — Guide the AI's overall approach without changing the questions. For example: "Focus especially on emotional reactions" or "Participants are power users, so skip basic usage questions if they demonstrate familiarity."
Plan for 2-3 minutes per question-probe pair — This accounts for the main question, the participant's response, a possible probe, and transitions.
Pilot test with AI Participants — Run your interview with AI Participants first to see how the AI handles your probes and whether the interview fits within the time limit.

Related Guides

Understanding Interview Types — Compare all three interview modes
Structured Interviews — For maximum consistency
Unstructured Interviews — For maximum exploration