By the end of the lesson, you can run a short, ethical, non-leading research conversation. You can separate the participant's quote, the observable behavior, and your own interpretation. You understand why it is important to ask about past, concrete situations.
Why is it hard to ask well?
At first glance an interview is simple: you sit down with someone and talk. In practice, three things spoil it most often. People are polite, so they say what they think you want to hear. People predict their own future behavior badly ("yes, I would use it"). And we ourselves tend to suggest the answer with the question. Interview technique is a toolbox built against these.
A conversation, not an interrogation
A user interview is the form of research in which the researcher asks a participant about their experiences, habits, and decisions. A semi-structured interview is built on predetermined topics but follows the natural flow of the conversation: there is a guide, but no fixed list of questions.
The power of the past example
Instead of general opinions, ask for a past, concrete example. After the question "Do you usually book appointments online?" can come: "Could you tell me about the last time? What started it? What did you do first? Where were you uncertain?" The details (open tabs, screenshots, an acquaintance asked, an abandoned step) often show more than a summarizing judgment.
Why does this work? Because a person recalls an event that happened, but invents a future intention. Recall can still be distorted, but substantially less than guessing.
The structure of the interview
Introduction and consent: say what the conversation is about, that there is no right or wrong answer, that they can stop at any time, and ask for consent to take notes.
Warm-up: easy questions about the participant's background and relationship to the task.
Main part: recalling the most recent concrete occasion, with detailed follow-up.
Closing: "Is there anything I did not ask but that would be important?" and thanks.
Questioning technique
A leading question is a question that suggests the desired answer. "Was it annoying that cancellation was so hidden?" already conveys what the participant should feel. The neutral version: "If you wanted to change the booking, where would you look for it?"
A few rules:
Use open questions: "What happened?", "How did you decide?", "What would you have expected here?"
Avoid compound questions: "Was it easy and fast?" is two questions in one; the respondent may feel yes for one and no for the other.
Use "why?" carefully: by itself it sometimes sounds accusatory. Better: "What was important to you in this?" or "What did you choose this based on?"
Follow interesting details: if the participant says something surprising, ask about it: "Could you tell me more about this?"
Do not argue, correct, or explain the system. If the participant misunderstands something, that is itself the data.
Use silence: if you wait for the next sentence, another important detail often comes.
Observation and thinking aloud
In observational research you do not just listen but also watch what the participant does in a task. If someone shows how they look for a program, you see which information they look for first, what they click before deciding, when they stop or go back. This often shows something different from what they tell afterwards.
Thinking aloud (think aloud) is a method in which the participant states their expectations and interpretations during the task. You can ask: "Tell me what you are looking for or what you expect!" But do not demand continuous speech: the task and the observation are more important. Later, in usability tests (chapter 8), we will use the same method.
Note-taking: quote, observation, interpretation
In notes, separate three kinds of elements, because mixing these up is the most common mistake in research.
Kind
Example
Why separate?
Quote
"I don't know if this is final now."
Verbatim evidence
Observation
The participant returned to the summary page twice
Verifiable behavior
Interpretation
The language of the confirmation did not give enough certainty
Your conclusion, a hypothesis, not fact
Interpretation is not bad, but it has to be marked separately, because later it must be confirmed or rejected with more data.
Ethics and data handling
Do not ask for unnecessary personal, health, financial, or other sensitive data. If screen sharing takes place, the participant can stop it at any time, and do not ask them to show a live password or private message. Even for short notes, mark who has access and how long you keep them. If you record, do so only with explicit permission. In research, the participant's time and trust are a resource, not free test material.
Worked example: an interview for Mai program
A participant, Réka, tells how they looked for a program last Friday. The researcher does not ask "Would a filter be good?", but: "Could you tell me how your Friday evening started? When did you begin searching?"
Réka says they looked at events on five tabs, and copied the time from one page to a friend. When the researcher asks what the most uncertain moment was, Réka says: "I didn't know whether I had to register in advance, and only on site did it turn out that I should have."
The note looks like this:
Quote: "Only on site did it turn out that I should have registered."
Observation: worked with five tabs, copied the time by hand.
Interpretation (hypothesis): the registration requirement is not visible early enough; to be checked with further interviews.
The researcher did not propose a solution here (for example "Would a marker on the card be good?"). The solution is a later step.
Common misconceptions
Claim
Clarification
"An interview is good if the participant talks a lot about my idea."
The goal is to understand the participant's situation, not to comment on the idea.
"If the participant says it's a good idea, that is evidence."
A polite opinion is not behavioral evidence.
"I should help if they get stuck."
The goal of the research is observation. Getting stuck is data.
In brief
Ask about past, concrete situations, not future intent.
Avoid leading and compound questions; use open, neutral wording.
Watch behavior too, because it often differs from what is said.
In the notes, always keep the quote, the observation, and the interpretation separate.
Protect the participant's data and trust.
Review questions
Why is it more useful to ask about a past example than about a future intention?
Rewrite a leading question as a neutral question.
Which is fact and which is interpretation: "The participant searched for two minutes"; "the navigation is bad"?
What personal data do you not need to collect for your project?
Interviews and observation
Objectives
By the end of the lesson, you can run a short, ethical, non-leading research conversation. You can separate the participant's quote, the observable behavior, and your own interpretation. You understand why it is important to ask about past, concrete situations.
Why is it hard to ask well?
At first glance an interview is simple: you sit down with someone and talk. In practice, three things spoil it most often. People are polite, so they say what they think you want to hear. People predict their own future behavior badly ("yes, I would use it"). And we ourselves tend to suggest the answer with the question. Interview technique is a toolbox built against these.
A conversation, not an interrogation
A user interview is the form of research in which the researcher asks a participant about their experiences, habits, and decisions. A semi-structured interview is built on predetermined topics but follows the natural flow of the conversation: there is a guide, but no fixed list of questions.
The power of the past example
Instead of general opinions, ask for a past, concrete example. After the question "Do you usually book appointments online?" can come: "Could you tell me about the last time? What started it? What did you do first? Where were you uncertain?" The details (open tabs, screenshots, an acquaintance asked, an abandoned step) often show more than a summarizing judgment.
Why does this work? Because a person recalls an event that happened, but invents a future intention. Recall can still be distorted, but substantially less than guessing.
The structure of the interview
Questioning technique
A leading question is a question that suggests the desired answer. "Was it annoying that cancellation was so hidden?" already conveys what the participant should feel. The neutral version: "If you wanted to change the booking, where would you look for it?"
A few rules:
Observation and thinking aloud
In observational research you do not just listen but also watch what the participant does in a task. If someone shows how they look for a program, you see which information they look for first, what they click before deciding, when they stop or go back. This often shows something different from what they tell afterwards.
Thinking aloud (think aloud) is a method in which the participant states their expectations and interpretations during the task. You can ask: "Tell me what you are looking for or what you expect!" But do not demand continuous speech: the task and the observation are more important. Later, in usability tests (chapter 8), we will use the same method.
Note-taking: quote, observation, interpretation
In notes, separate three kinds of elements, because mixing these up is the most common mistake in research.
Interpretation is not bad, but it has to be marked separately, because later it must be confirmed or rejected with more data.
Ethics and data handling
Do not ask for unnecessary personal, health, financial, or other sensitive data. If screen sharing takes place, the participant can stop it at any time, and do not ask them to show a live password or private message. Even for short notes, mark who has access and how long you keep them. If you record, do so only with explicit permission. In research, the participant's time and trust are a resource, not free test material.
Worked example: an interview for Mai program
A participant, Réka, tells how they looked for a program last Friday. The researcher does not ask "Would a filter be good?", but: "Could you tell me how your Friday evening started? When did you begin searching?"
Réka says they looked at events on five tabs, and copied the time from one page to a friend. When the researcher asks what the most uncertain moment was, Réka says: "I didn't know whether I had to register in advance, and only on site did it turn out that I should have."
The note looks like this:
The researcher did not propose a solution here (for example "Would a marker on the card be good?"). The solution is a later step.
Common misconceptions
In brief
Review questions