The user interview is one of the most widely used techniques in UX research. This direct exchange with a user consists of holding a conversation whose topics are defined beforehand in an interview guide.
Conducting an interview with a user helps you understand the motivations hidden behind observable behaviours. This logic of answering the “why?” of an observed situation is the basis of qualitative approaches in UX.
Contrary to popular belief, running an in-depth user interview cannot be improvised. It draws on techniques you learn by practising, and it requires care in preparing your interviews with users.
01
The vox pop
short exchanges (5 to 15 min) with users in their natural context, without prior recruitment. Ideal for quick exploration or light validation.
02
The semi-structured interview
the most common format. A guide of questions serves as a common thread, but the researcher adapts to the participant's discourse. A good balance between structure and freedom of exploration.
03
The structured interview
questions set in advance, asked in the same order to all participants. More comparable across interviews, but less rich in unexpected data.
04
The explicitation interview
you ask the participant to go back in detail over a past action or situation. It reveals implicit steps and rationalisations the participant would not have spontaneously articulated.
05
The self-confrontation interview
you replay a recording of the participant's own session in front of them and ask them to explain what they were doing and thinking at that moment. The most powerful method for reaching cognitive processes in a real context.
1. Introduction: You start by introducing yourself and putting the participant at ease. You recall the purpose of the exchange and explain the different themes that will be covered. Importantly, you inform the participant about the use that will be made of the collected data (GDPR procedure) and make sure you have their consent to take part freely in the interview;
2. Setting the scene: You find out about the participant to know who you are speaking to and where they are coming from;
3. Running the interview: You cover each of the themes and questions from the guide prepared beforehand. You avoid doing so too linearly, adapting to the order in which the user wants to share their feedback. You listen actively, you don’t fill the silences, and you follow up intelligently;
4. Retrospective: You go back over what was said with the user to complete it and avoid any misinterpretation;
5. Debriefing: You make sure to wrap up properly, you leave room to add anything that wasn’t covered, and you thank the participant.
In user research, semi-structured interviews are most often the preferred choice. It is often the right one, though not always the only one.
When exploring a very specific subject, such as an operator’s work procedures for example, the explicitation interview gives better results. You follow the participant’s account, not your script.
When you have precise hypotheses to compare across participants, the structured format is more suitable. And when you want to understand why someone did something in their real context, the self-confrontation interview, by replaying a recorded session, lets you look into the cognitive processes the user relied on.
There's no magic answer to this question. The only guarantee is that a single interview is better than none at all. In qualitative approaches, we talk about saturation, the point from which you no longer learn anything new. When the profiles are homogeneous (e.g. a group of users of one service), you get very rich results even with fewer than ten users involved. It all depends on your context!
The real risk in an interview, whatever the type, is social desirability. In a test context, participants tend to say what they think is expected of them, or to rationalise behaviours they hadn't really deliberated. It's to avoid this bias that we refrain from asking someone what they think. We favour recalling a recent situation, along the lines of « tell me about the last time you tried to... ». This gives access to a concrete lived experience, not an opinion reconstructed after the fact.
Depending on the interview format and the type of data recorded during a user interview, the analysis can vary. Very generally, you start by gathering the participants' answers across the different themes discussed. From this raw material, you can then begin to look for whether themes recur and see how the participants interviewed break down according to their answers. When the interviews are transcribed, you can also carry out a discourse analysis. For each UX project the goals vary, so the principle of the analysis is to confront the material collected with the hypotheses and/or the points that interest us most for a study.
A successful interview relies on good preparation. Both on creating the interview outline, with the choice of themes, and on the format you offer for answering. You can for instance draw on sharing positive or negative anecdotes to get people talking about a lived experience. You can also imagine offering answers in the form of a top 3 of situations encountered, or asking what they would do if they had a magic wand. The aim is to lead the user to share as much detail as possible without feeling they are being interrogated. It's the art of questioning.
The follow-up question is very important for a one-to-one interview. Once you've defined the themes to discuss with a user, you list specific follow-up questions. These questions are designed to help steer the discussion towards sub-themes smoothly during the exchange. During the interview, they let you carry on after an answer, ask for more detail, ask why a given answer, and so on.
A discussion group, run by a facilitator, where users speak about a shared theme through structured debate.
Read more about Focus groupPersonas come out of work on user-profile modelling. They highlight behavioural archetypes that prove useful when designing.
Read more about PersonaEveryone talks about user experience, service design, ergonomics...
It's not always clear, and you'd like to find out, or sharpen what you already know.
