5 Steps to Create Good User Interview Questions By @Metacole — A Comprehensive Guide

Teo Yu Sheng
Interactive Mind
Published in
9 min readAug 22, 2015


I give free mentorship on ADPList: Book a session here if you’re interested!

Coleman Yee is an experienced UX practitioner, and over a month ago I’ve had the chance of personally learning from him. This post is a crystallisation of some of the lessons that I’ve picked up from Coleman.

User interviews are an exceptionally useful tool for user research, because it allows you to speak directly to users, and get responses to specific questions that you have. If done well, it could also uncover nuances or directions that were previously unknown or unthought of.

Like most research, however, user interviews are prone to bias, and interpreting responses can be tricky. But with a good guide in hand — and lots of practice — we can avoid the most common pitfalls and get the most out of user interviews.

The first step in conducting user interviews is the creation of interview questions. Badly scripted interview questions can not only nullify the benefits of user research, but also lead product development down the wrong path.

Here are 5 comprehensive steps that will help you create a list of good user interview questions.

1. Know what you want answered.

This might sound obvious, but it’s important that you know why you’re conducting user interviews. In other words, you need to know what questions you want answered. These are the problem statements — write them down in a doc (or on a piece of paper).

Examples of Problems Statements:

“Why do people make purchases online?”

“How do people make purchases online?”

“What’s the difference between online and offline purchasing for the consumer?”

2. Reframe the problem statements.

Now, think of different perspectives to express the same problem statements. This is more than an exercise in paraphrasing.

You’d have to imagine each statement as a piece of gem: your job is to turn it in your hand and look at it from different angles.

Here’s an example:


“Why do people make purchases online?”

Reframe 1:

“What makes people want to buy online?”

Reframes 2 & 3:

“What makes a product saleable online?”

“What makes a product unsaleable online?”

As you can see, each of these reframes provide a different perspective to the same topic: the original statement is more logic/rationale-driven, while reframe 1 made it more emotion/desire-driven, and the last 2 reframes focus a bit more on the product instead of the consumers.

The benefits of reframing problem statements are two-fold:

  1. You stumble upon questions/perspectives that you might not have considered at the beginning, that might help you with your research.
  2. You lay a base on which to build your interview questions.

3. Build a list of questions.

If you did step 2 above properly, you’d already have laid a good foundation to build your questions. In other words, you kinda know what questions you want to ask. The tricky bit, then, is how you should ask them.

Here are some general guidelines to help you phrase your questions properly:

A) Don’t ask leading questions.

Leading questions are questions that are framed (wittingly or unwittingly) in a way that influences the answers given by respondents.

Leading Question:

“How angry do you usually feel when an online transaction fails to go through successfully?”

The example above is leading because it focuses the respondent’s mind to a single emotion (anger), and makes it unlikely that he/she will talk about other emotions, like anxiety, or shock.

The purpose of conducting an interview is to get real, truthful answers to questions that you have. Asking leading questions defeats that purpose, so don’t.

Non-leading Question:

“Recall a time when an online transaction failed to go through successfully. How did you feel?”

Unlike the first example, this non-leading question opens up the whole range of emotions to the respondent.

Note, however, that the question still frames the respondent to only think about the emotions that he/she associates with failed online transactions.

Unless the research that you’re doing is concerned primarily about the emotions of online shoppers, you might be better off asking this:

Even More Non-leading Question:

“Recall a time when an online transaction failed to go through successfully. What did you think or feel?”

Now, the question isn’t restricted only to emotions, but also includes the respondent’s thoughts. Accordingly, you should probe further about their emotions if they focus too much on their thoughts, and vice versa, to get both aspects of the answer.

All in all, it’s easy to ask leading questions, but also easy to avoid them. All it takes is a little more thought.

B) Ask about specific incidents in the past, when possible.

The more speculative a question is, the less you can rely on it. That’s because you’re effectively asking your respondents to create scenarios in their heads, and then answer questions based on those imagined circumstances.

In other words, the rigour — and truthfulness — of their answers become dependent on the rigour and truthfulness of their imagination; you’re increasing your chances of getting bad data.

Using the same when-an-online-transaction-fails question, here’s an example of a speculative, versus a context-based, variation:

Speculative Question Based on Imagination:

“Tell me what goes on in your head when an online transaction fails to go through successfully.”

Context-based Question Grounded in Reality:

“Tell me what went on in your head the last time an online transaction fails to go through successfully.”

The second question gives respondents a specific context to hold on to when answering your question, which in turn increases your chance of getting genuine, insightful data.

Of course, speculative questions have their advantages, but as much as possible, they shouldn’t be used to uncover current or past behaviours.

C) Ask more open-ended questions.

Sometimes, you might want to cast your net a little wider, because the respondent might have something relevant to say that you didn’t think of asking.

Specific Question:

“What was the last thing you bought online?”

Open-ended Question:

“Tell me more about the last time you purchased something online.”

The first question asks specifically about the product being bought, while the latter asks about the experience in general. If there were any part of the purchase that stood out to the respondent, it would’ve been missed by the first question, but captured by the second.

And if the respondent didn’t mention anything about their purchase that you think is important, you can always follow up and ask.

D) You can ask more than one question to find out the same thing.

Doing so allows you to have more than one data point to verify that you’ve understood the respondent correctly, and to check for contradictions in the respondent’s answers. The reframing exercise in Step 2 should’ve already yielded you multiple questions for the same topic.

Here’s an example:

Question 1:

“What are some reasons why you shop online?”

Question 2:

“What are some things that will make you decide not to buy from an online store?”

Do note that these 2 questions don’t necessarily have to come one after the other; pick a flow of questions that make the most sense to the respondent (more on that in Step 5C).

(Extra reading: if you’re interested, you can look up a similar research method called “data triangulation”, which uses multiple data, theories, or methods to inform research findings.)

E) Don’t ask if they’d buy your product.

As tempting as it is, don’t.

As I mentioned in this blogpost, asking a respondent if they’d buy your product is akin to asking a mother of two if she loves one of her kids more — it’ll take a brutally honest person to say no.

What you should do, instead, is to ask questions that reveal your respondents’ intent to purchase.

4. Paraphrase each question 2–3 times.

The point of this step is to have a list of paraphrases, in case the respondent doesn’t understand what you mean the first time you ask them.

You can skip this step for super-obvious questions, like what the respondent usually shops for online.

5. Give the whole list of questions a structure.

A user interview is usually made up of 4 parts: an introduction, some warm up questions, the main body questions, and a wrap up.

A) Introduction

The point of this section is to give the respondent some idea of what’s going on, so that they don’t feel too confused throughout the interview. But be sure not to tell them too much, because you don’t want to overly frame their minds to think in a specific way.

For instance, let’s say you’re working on an app that secures online transactions, and you want to find out how important online transaction security is to online shoppers.

If you let your respondents know exactly what you’re working on, you would’ve already preconditioned them to think about online security. This makes it hard for you to determine if online security is really something that they think about when they shop online.

In general, it’s better to keep the introduction slightly vague, but let your respondents know that you’ll address their queries at the end of the interview.

This is also a good point to let the respondent sign any release form that you have, and ask for permission to record the audio/video of the interview.

Here’s an example of an introduction that I often use:

“Hi Alice, thanks for coming down. My name is ______, and I’ll be conducting the interview today.

We’re doing some market research, so some of the questions might be a little personal, and will concern your lifestyle. If you feel uncomfortable answering any of the questions, just let me know and we can skip it. I can’t tell you much about what we do or the exact reason we’re doing this research now, because it might affect the way you answer the questions. But I can definitely tell you after the interview is over.

Do you have any questions so far?
[pause for respondent’s reply]

If you don’t mind, I would record the audio for this interview; it will only be used internally, for the purposes of this research.”

B) Warm up questions

At this point, you want to start getting your respondents comfortable talking. Ask them 3–5 generic questions that are related to the topic of your interview.

Examples of warm up questions:

“What’s your occupation / what do you do all day?”

“Can you tell me more about your hobbies?”

“How often do you use the internet? etc.”

C) Main body questions

This is the meat of the interview, and pretty much the whole point of conducting one to begin with.

You should already have a full list of questions that you want to ask your respondents, and the last thing you’ll have to do now is to order them.

As much as possible:

  1. Start with questions relating to specific incidents in the past, before moving on to speculative ones.
  2. Follow a flow that makes sense to the respondents. For example, ask about their general buying behaviours before asking specific questions about their thought process behind different stages of the buying process.

D) Wrap up

At the end of the interview, you should give your respondent a sense of closure.

Don’t forget to ask them if they have anything they’d like to add or ask — more than once, I’ve received insightful data from respondents at the end of their interview.

Finally, thank them for their time and contribution towards your research.


5 steps to create good user interview questions:

1) Come up with problem statements.

2) Reframe your problem statements.

3) Build a list of questions based on your reframed problem statements.

4) Paraphrase each question 2–3 times.

5) Structure your questions properly.

Found this article helpful? Recommend it! =)



Teo Yu Sheng
Interactive Mind

Founder of queer brand Heckin’ Unicorn. Previously PD & PM @ Carousell & ReferralCandy. I code in my free time. www.yushengteo.com