Probing Questions in Qualitative Interviews: 4 Types

Last updated: November 7, 2026

The interview guide gets most of the attention in a methods section, but the depth of qualitative data is usually decided somewhere else: in the follow-up questions nobody wrote down in advance. Probing questions in qualitative interviews are what turn "it was fine" into a story you can analyze. This guide defines probing, explains the four probe types in Robinson's 2023 framework, gives example wording for each, and covers what changes when an AI moderator does the probing.

It is written for PhD students, research labs and qualitative or market research firms. UX and customer-insights teams can use the same toolkit.

Key takeaways

• A probe is a follow-up to a specific answer, not a new topic. It asks the participant to clarify, elaborate, illustrate or explain what they just said.
• Four probe types cover most needs: descriptive detail, idiographic memory, clarifying and explanatory (Robinson, 2023).
• Neutral wording matters more than clever wording. A probe that suggests an answer ("Was that frustrating?") is a leading question.
• Probe the participant's words, not your hypothesis. Reuse the term they used and ask what it means to them.
• AI moderators can probe, but depth varies. A 2026 experience report on AI-conducted interviews lists "insufficient depth" and "limited sensitivity to answers" among its observed limitations, so test the follow-ups before fieldwork.

What are probing questions in qualitative research?

Probing questions are follow-up prompts that ask a participant to clarify, elaborate, illustrate or explain a previous answer. Robinson (2023) uses close to this definition in "Probing in qualitative research interviews: theory and practice" (Qualitative Research in Psychology, doi 10.1080/14780887.2023.2238625). A probe always points backward to something the participant has said. A main question opens a topic; a probe deepens it.

Probing is what separates a semi-structured interview from a spoken questionnaire. In a fully structured interview, every participant hears the same questions in the same order and the interviewer does not improvise. In semi-structured and in-depth interviews, the guide sets the topics and the interviewer decides, answer by answer, whether to probe or move on.

Probes vs. follow-up questions vs. prompts

The terms overlap, and textbooks use them differently. A practical distinction:

• Main question: written in the guide, opens a topic ("Tell me about the last time you analyzed interview data.").
• Probe: short, generic in form, asks for more of what was just said ("Can you give me an example?").
• Follow-up question: content-specific, picks up a new thread the participant raised ("You mentioned your supervisor reviewed the codes. How did that work?").
• Prompt: a cue written into the guide in case the participant does not mention something ("If not mentioned: ask about software used.").

The four types of probing questions (DICE)

Robinson (2023) proposes a typology summarized as DICE: Descriptive detail probes, Idiographic memory probes, Clarifying probes and Explanatory probes. The paper grounds it in narrative, self-disclosure, autobiographical memory and attribution theory. A CHI 2025 study by Jacobsen, Cox, Griggio and van Berkel (arXiv:2503.08582) later tested these four probes in a survey chatbot with 64 participants and reported that probes improved data quality, with different probes working best at different research stages.

1. Descriptive detail probes

These ask for more sensory, practical or contextual detail about what happened. Use them when an answer is accurate but thin.

• "What did that look like?"
• "Walk me through what you did next."
• "Who else was there?"

2. Idiographic memory probes

These move a participant from generalities ("I usually...") to one specific, remembered occasion. Answers about a single real episode tend to be more concrete and less rehearsed than answers about habits.

• "Can you think of a particular time that happened? Tell me about it."
• "When was the most recent time?"
• "What happened on that day?"

3. Clarifying probes

These check meaning. Participants use words like "messy," "fair" or "real research" that sound clear but mean different things to different people. A clarifying probe reuses the participant's term instead of substituting your own.

• "When you say 'messy,' what do you mean?"
• "Can you say a bit more about 'real research'?"
• "I want to make sure I understand. Was that before or after the deadline?"

4. Explanatory probes

These ask the participant to account for something: why they acted, what they think caused an outcome, how they make sense of it. Ask them after you have the description, not before, so the explanation is anchored to an event.

• "What made you decide to do it that way?"
• "Why do you think that happened?"
• "How did that change the way you work now?"

Other probing techniques worth knowing

• Silent probe: a pause of a few seconds after an answer. Participants often fill it with more detail.
• Echo probe: repeating the participant's last key phrase as a question ("Too late?").
• Contrast probe: "How was that different from the previous project?"
• Laddering: a chain of explanatory probes that moves from attributes to consequences to values. See our laddering guide.

Probe Decision Matrix: when to probe and which probe to use

The hardest skill is not knowing probe wording but deciding, in real time, which one an answer needs. This matrix is a Qualitati-original tool for training interviewers and for writing probing instructions into an interview guide.

Signal in the answer	Example answer	Probe type	Neutral probe	Avoid
Short or vague	"It was fine."	Descriptive detail	"Tell me more about how it went."	"So you were satisfied?"
Generalization	"I always check the codes twice."	Idiographic memory	"Can you tell me about the last time you did that?"	"That's good practice, right?"
Ambiguous or loaded term	"The process felt unfair."	Clarifying	"What does 'unfair' mean to you here?"	"Unfair because of your manager?"
Decision or outcome with no reason	"So I switched tools."	Explanatory	"What led you to switch?"	"Was it the price?"
Emotion or hesitation	"Well... it was a lot."	Descriptive, then explanatory	"What was that like for you?"	"That must have been stressful."
Contradiction with an earlier answer	"I never use AI." (earlier: "ChatGPT helps me draft")	Clarifying	"Earlier you mentioned drafting with ChatGPT. How does that fit?"	"You just contradicted yourself."
Rich, complete answer on a covered topic	A detailed story with examples	None	Move to the next guide question	Probing for the sake of it

Probe quality checklist

• Is it one question, not two joined by "and"?
• Is it open (how, what, tell me about) rather than yes/no?
• Does it reuse the participant's own words?
• Does it avoid suggesting a feeling, reason or answer?
• Does it stay on the participant's topic rather than introducing yours?
• Is it shorter than the participant's answer?

How to write probes into an interview guide

Most probing is improvised, but planned probes make interviews more consistent across interviewers and easier to audit. Under each main question, list two or three probes and a short note on what you are trying to learn. For example:

• Main question: "Tell me about the last time you coded a set of interviews."
• Probes: "What did you do first?" (descriptive) · "What made you choose that codebook?" (explanatory) · "When you say 'saturated,' what did that look like?" (clarifying)
• Purpose note: understand where researchers lose time and why.

Pilot the guide with one or two practice interviews and check whether the probes produce new information or only repetition. Our guide to pilot-testing an interview guide covers this step, and our research question examples help connect probes back to what the study is meant to answer.

Probing questions in AI-moderated interviews

AI moderators now probe in real time, and the research on them is growing quickly. Wuttke, Lang, Klamm, Würschinger and Kreuter (arXiv:2606.20064, June 18, 2026) compared AI-led voice, chat and free-choice interviews with a standardized survey among 571 respondents and found that completers rated the AI interview at or above the survey. The transcripts surfaced reasoning the survey did not capture.

Depth is still the weak point. In an experience report on AI-conducted interviews in two software engineering studies (66 submissions), Gheyi, Albuquerque, Ribeiro and Perkusich (arXiv:2607.14452, July 16, 2026) list generic questions, limited sensitivity to answers and insufficient depth among the limitations they observed. The authors also state their results do not establish equivalence to human-conducted interviews.

The practical lesson: an AI moderator probes only as well as its instructions and its model allow. Write probe instructions into the guide, pilot with real or simulated participants, and read the transcripts for the signals in the matrix above. For human interviewers who want help in the moment, see our post on real-time AI probing suggestions.

Where Qualitati fits

Qualitati is a European qualitative research platform for universities and research companies, with AI-moderated interviews in text and voice, conversational surveys with AI follow-ups, and analysis tools such as ThemeLens and QDA Workspace. Probing is part of how the AI interviewer is designed to work:

• Probing instructions in the guide. Researchers write main questions and probing instructions into the interview guide, and can request an AI quality score for the guide that includes a probes dimension.
• Neutral, one-at-a-time follow-ups. In flexible mode, the interviewer is instructed to ask one open, neutral question per turn and to follow up on brief, vague, emotional or surprising answers using the participant's own words.
• A supervisor check. A second model reviews the conversation and flags missed probes, leading questions and contradictions so the interviewer can adjust.
• Structured mode when you need it. If your design requires every participant to hear the same questions, structured mode sticks to the outline and only uses probes you wrote in advance.
• Active Listener for human interviewers. A human runs the interview and receives real-time prompts and section tracking.

Interviews run in 10 languages, and the free tier includes 30 monthly credits with no credit card required. See pricing for current plans.

Limitations and methodology concerns

• Probing shapes the data. Two interviewers who probe differently will produce different transcripts from similar participants. Report your probing approach in the methods section.
• More probes are not always better. Repeated probing can feel like interrogation, especially on sensitive topics. Respect a participant's choice not to elaborate.
• Explanatory probes invite rationalization. People will give a reason when asked "why," even if they did not reason that way at the time. Treat answers as accounts, not causal evidence.
• AI probing needs human review. The studies cited above are early, mostly preprints, and do not show that AI probing matches skilled human interviewers. A researcher should read transcripts before relying on them.

Who this is for, and when not to use it

This toolkit fits semi-structured and in-depth interviews, focus groups and conversational surveys with open follow-ups. It is not the right approach for standardized survey interviews where comparability requires identical wording for every respondent, or for cognitive interviews that follow a fixed probing protocol.

Frequently asked questions

What is a probing question in an interview?

A probing question is a short follow-up that asks a participant to clarify, elaborate, illustrate or explain something they just said. It deepens an existing answer rather than opening a new topic.

What are the four types of probes?

Robinson (2023) describes four: descriptive detail probes, idiographic memory probes (asking about one specific occasion), clarifying probes and explanatory probes. The set is often abbreviated as DICE.

What is the difference between a probe and a leading question?

A probe asks for more without suggesting the answer ("What was that like?"). A leading question suggests a feeling, reason or answer ("Was that frustrating?"), which can bias what the participant says next.

How many probes should I ask per question?

There is no fixed number. Probe until the answer is concrete and you understand what the participant means, then move on. Plan two or three probes per main question and use them only when the answer needs them.

Can AI interviewers ask good probing questions?

They can ask relevant follow-ups, and a 2026 study with 571 respondents found completers rated AI interviews at or above a standard survey. Reported weaknesses include generic questions and insufficient depth, so pilot the guide and review transcripts.

Bottom line

Probing questions in qualitative interviews are where depth comes from. Learn the four DICE types, use the decision matrix to match the probe to the answer, keep every probe neutral and short, and write your probing approach into both the guide and the methods section. Whether a person or an AI asks the questions, the same rules apply.

Want to see probing in practice? Start free with 30 credits and run an AI-moderated interview with your own guide, or read our discussion guide template for AI moderators.

Last updated: November 7, 2026

Study findings are summarized from the public abstracts and pages linked above as of November 7, 2026. Qualitati is not affiliated with the authors.