Hey Juno

How to ask probing questions in qualitative interviews

By Hey Juno · Published · Updated

A probing question is a follow-up that asks a person to say more about something they have already said. In qualitative interviews, the first answer is usually a summary, so the useful material tends to arrive on the second, third or fourth question. Good probing stays with the person's own words, asks for specifics, and knows when a thread has stopped paying off.

This guide covers the main kinds of follow-up questions in interviews, how to use them without leading, how to decide when to stop, and how Juno, the AI researcher from Hey Juno, probes while keeping a study's coverage intact.

Why the first answer is rarely the full answer

Ask someone how their last holiday went and you will hear "great, really relaxing". That is a verdict, not an experience. People compress what happened into a label because it is polite, quick and usually enough in conversation.

For research, the label is where you start. It hides the details that explain it: what happened, in what order, what the person compared it with, and what they left out because it seemed too small to mention. Rehearsed or generalised answers also sound more confident than they are. "I always compare prices first" is easy to say. What the person did on Tuesday is harder to invent and more useful to hear.

Probing is how you get from the label to the experience. It is not about pushing harder. It is about showing that you want the detail, and then following it.

Types of probes

These are illustrative examples from everyday topics. They show the technique, not findings from any study.

Clarifying probes. Check what a word means to the person. "You said the commute was 'a nightmare'. What does that look like on a normal morning?" Words like fine, easy, expensive and stressful carry different meanings for different people. Never assume yours.

Elaborating probes. Invite more without steering. "Tell me more about that." "What else comes to mind?" These are the plainest probes and often the best, because they add nothing of your own.

Example probes. Ask for a specific case. "Can you think of a time that happened?" A concrete story shows how the person actually behaves. It also separates what they do from what they think they ought to do.

Contrast probes. Ask what the thing is measured against. "You said the new gym is better. Better than what?" Contrast shows the standard behind the judgement, which is often the most useful finding.

The "last time" anchor. Move from a general habit to the most recent real event. "When did you last cook a meal from scratch? Walk me through it." Recent events are remembered more accurately than typical ones, and they carry detail that generalities smooth over. Questions about real events also avoid asking people to predict their own behaviour, which they do poorly.

Silence and echoing. Pause for a moment after an answer. People often keep talking to fill the gap, and what they add is frequently the real point. Echoing means repeating the last few words as a question: "Nobody told you?" It signals attention and hands the thread straight back.

Laddering: following a reason up a level

Laddering is a structured way to probe. You start with something the person did or noticed, then keep asking why it matters until you reach a value or motivation. Reynolds and Gutman described the technique for consumer research in 1988 (see Sources). In plain terms: "I chose the lighter backpack." Why does lighter matter? "My shoulder aches." What would that change? "I'd actually walk to work instead of driving."

Use it lightly. Asking "why" over and over can feel like an interrogation, and some people will invent a reason just to satisfy the question. Alternate it with the other probes, and stop when the answers stop being concrete.

How to probe without leading

A probe can bias an answer as easily as an opening question can. The rules are the same ones in our guide to leading questions, so this section only covers what is specific to follow-ups.

  • Use their words, not yours. If the person said "chaotic", ask about chaotic. Swapping in "stressful" or "disorganised" plants your interpretation.

  • Reflect without interpreting. "You mentioned the form took a while" is a reflection. "So the form was too long?" is a conclusion offered for agreement.

  • Ask for both sides. If you probe only the complaints, you hear only complaints.

  • Keep warmth neutral. Enthusiasm after one kind of answer teaches people which answers please you.

  • Offer options last. Ask openly first. Give examples only if the person is stuck, and then give several.

If you want the full list of what to avoid when writing questions, read that guide first and come back here.

When to stop probing a thread

Probing has a cost. Each follow-up uses the participant's time and patience, and the interview has other ground to cover. Move on when:

  • The answers repeat. You are hearing the same point in new words.

  • The detail stops being concrete. Examples have turned into generalities again.

  • The person signals discomfort. Shorter answers, a change of tone, or a polite refusal. Respect it and take another route, or move on.

  • The thread no longer bears on the decision. It may be interesting, but the study exists to inform something specific.

  • Time is short. Other topics still need their share.

Moving on is not abandoning the thread. If something important came up, you can return to it later from a different angle. A deliberate return is fine. Asking the same question again by accident is not.

How Juno probes adaptively and keeps the study's coverage

Juno works from research objectives rather than a fixed list of questions, so it can follow a participant somewhere unplanned. It decides when to probe by weighing how relevant the answer is, how concrete it was, what is still uncovered and how comfortable the participant seems. It digs where a thread is earning something and moves on when it is not.

Several things stay fixed while the follow-ups change:

  • Coverage. Juno maps the territory breadth-first before going deep on one thread, so an early tangent does not crowd out ground that still matters.

  • Vague answers. Juno asks for a concrete example, covering what happened, when, who was involved and what followed, rather than accepting a polite abstraction.

  • Wording that must not change. If a team marks a question as verbatim, Juno asks it word for word, then returns to adaptive follow-ups around it.

  • Question checks. Questions are checked before they are asked for leading phrasing, embedded answers, false either/or choices, fed examples and presuppositions.

  • Repetition. Juno tracks what has been covered so that returning to a topic is deliberate.

You can read every turn of every interview to judge the probing yourself. The full list of interviewing craft is on the competencies page, and the interview page shows a worked exchange.

Where this doesn't apply

Probing suits open, exploratory interviews. In a structured survey or a standardised questionnaire, interviewers are meant to ask every question identically, and ad hoc follow-ups would break comparability. Probing is also the wrong tool when someone is distressed or has said they would rather not discuss a topic. Careful interviewing there means letting them decline. The examples above are illustrative and will not fit every culture or setting, so adapt the tone to the people you are speaking with.

Frequently asked questions

How many follow-up questions should I ask on one topic?

There is no fixed number. Keep going while each answer adds something new and concrete, and stop when answers repeat or turn into generalities.

What is the difference between probing and leading?

A probe asks for more of what the person has already said, in their words. A leading question adds your own view or suggests the answer. The test is whether the person could answer without learning what you expect.

Is it better to ask "why" or "how"?

Often "how" and "what happened" work better. "Why" can sound like a demand to justify. Asking for a sequence of events tends to produce more detail and less defensiveness.

Can an AI interviewer probe as well as a person?

It can follow up on concrete detail, contradiction and unfinished thoughts, and it can keep coverage consistent across many interviews. Judge it on the transcripts. For more, see how AI-moderated and human-moderated interviews compare.

Sources

  • Reynolds, T. J. and Gutman, J. (1988). Laddering theory, method, analysis, and interpretation. Journal of Advertising Research, 28(1), 11-31 (February/March 1988).