Conducting Effective Discovery Interviews

Suppose you are asked to improve a system that someone uses every day. You could guess at what they need, or you could ask them to describe a real working session in detail. The second option is the starting point for a discovery interview. The source describes it as a structured conversation with a line user, meaning the person whose daily work the system is supposed to change. Its purpose is to learn how that work actually happens and where it creates friction, so that any later improvement rests on real practice rather than assumptions.

This article follows one session from beginning to end. You will see how the session is divided into stages, how the interviewer keeps two separate layers of record, how a reusable script is organized, and which mistakes tend to weaken the result. Once you can see the shape of one session, the reasons behind the method become easier to read.

The fixed, short length matters. Because the time is limited, the interviewer focuses on one concrete instance of the work rather than trying to cover everything the person does. A single real afternoon described in detail is worth more to the interviewer than a broad summary of an entire job.

Follow the session in order. It opens with a short introduction that sets the terms. The interviewer then moves to the core question, the centre of the interview, where the person narrates a specific past period. After that, follow-up prompts dig into the parts of that narrative that matter, and each relevant follow-up receives roughly five minutes. Throughout the session the interviewer keeps the spoken exchange and a set of private working notes, and the material that comes out of the session is recorded as notes.

Notice where the notes come in. They are not produced once at the end. The source notes that recording open questions during the session lets later follow-up build directly on what was learned. In other words, each stage feeds the next: what the person says in the core walkthrough determines which follow-ups are relevant, and the answers to those follow-ups are added to the record.

An interviewer keeps two layers of record. The first is the conversation itself, which records what was said. The second is the set of internal annotations, which are the interviewer's own working notes, kept separate from what the interviewee said. An annotation is written for the interviewer rather than for the person being interviewed. The conversation records what was said; the annotations record what the interviewer made of it.

Keeping these layers apart is useful. A raw transcript of questions and answers tells you what happened, but it does not tell you what matters. Annotations let you mark which answers point to a real step in the workflow, which sound like a workaround, and which raise a question to follow up on. Without that second layer, a thirty-minute conversation becomes a pile of statements that is hard to reason about afterward.

The reusable script organizes the interview into three parts: an opening, a core question, and follow-up questions. Each part has its own job, and the core question and the follow-ups work together as the body of the session. The script also sets practical conditions for the meeting: bring no slide deck, because the interview is not a presentation, and bring a notebook, because your job in the room is to record what the person says.

The opening is short, but it sets the terms of the whole conversation. It has three jobs: establishing who you are, making clear that you are there to learn rather than to sell, and agreeing on how the interview will run. An interviewee who expects a sales presentation will describe their work differently, often with more polish and less candor, so the purpose statement is not a formality.

The opening also addresses confidentiality. The script says that what the person shares will go to the sponsor only in summary form, and that it will not be attributed to them personally without asking first. Stating this early builds the trust the interviewee needs in order to speak candidly. The third job is ground rules. The script sets two: the interviewer asks about what the person actually did, not about what they wish were possible; and the interviewee is invited to push back when a question does not make sense, with the explicit preference that a confused answer is more useful than a polite one.

The core question is the centre of the interview and the part the script expects to take fifteen to twenty minutes. It is open-ended, which means it asks the person to narrate rather than to answer yes or no or to pick from options. The source describes the question as asking the interviewee to walk through what they did during a specific earlier period, from a clear starting point to a clear ending point.

Anchoring the request to a concrete, recent period is what makes it useful. A general question about how the person usually works invites a summary shaped by memory and habit. A bounded walkthrough of a real afternoon tends to surface the actual sequence of steps, including the detours and waiting that a summary would smooth over.

The script allows one or two clarifying questions when something needs explanation. A clarifying question asks for more detail about what the person has already said. The source gives the form such a clarification takes: asking what the person did specifically in a system they mentioned. The clarifying question picks up the person's own words and does not introduce new ones.

Once the core walkthrough has produced material, follow-up prompts help you dig into the parts that matter. Each follow-up has a rough allowance of about five minutes, and you use only those that are relevant to what the person has already described. The follow-ups contain placeholders such as a specific task, a colleague, or a system, so they must be tailored to the person's own words before you use them.

Throughout the session the interviewer records what the person said while keeping separate internal annotations about what it means. Each stage, from opening to core walkthrough to relevant follow-ups, feeds the recorded notes, so that the material the team later reasons from reflects the real work rather than assumptions about it.