Riddle me this: when does design – like, Designwith-a-capital-D Design – actually start? I’ll give you a hint: it’s not the moment you start pushing pixels around the screen.

The research that happens before committing pen to digital paper is as critical as – and in many ways more important than – the execution itself. Innovative and effective design doesn’t spontaneously occur in a vacuum, and it’s research that gives the design process the oxygen it needs to breathe and flourish.

Research gives the design process the oxygen it needs to flourish

Nonexistent, flimsy or even biased research insights (whether intentional or accidental) can have serious consequences down the line, including wasted time & effort, rejection of a feature or even entire products by users. The stakes are high, even at this early stage (anyone remember Digg 4.0’s disastrous rollout?), but before I’m accused of pointing fingers, I’ve got some launch disaster stories of my own – so I know intimately the pain caused by insights that are off the mark due to flawed research. (Remind me to write another post about that sometime, it’s a good story…)

In the research phase of any design, one skill that’s paramount is crafting good questions for your user interviews. Face-to-face interviews with your users and potential customers are one of the very best ways to get the deepest insights to inform your design, and the questions you ask must be carefully staged in order to get to them.

In this post, I’m going to share with you the techniques we use to counter our own biases in user interviews, and unearth the juicy nuggets & epiphanies that power the design work we do. By the time you’re done here, you’ll be ready to conduct an effective user interview that yields actionable insights you and your team can rely on to inform your design.

Already a research pro & just wanted a refresher? Skip to the good stuff below:

Common misconceptions about user interviews

Misconception 1: Interviewing is just talking

A user interview is not a social conversation, it’s a different, highly structured form of give-and-take communication that takes practice to master.

Close your eyes, and imagine an interrogation room in a police station. Can you see it? You might have envisioned flickering fluorescent lighting, an uncomfortable metal chair (complete with wobbly legs), and joyless, drab-colored walls.

Something like this, maybe?

Something like this, maybe?

Now, try imagining the set of a late night talk show. Perhaps you’re seeing a plush, comfy couch slightly angled next to a solid-looking wooden desk, both under bright, warm lighting. There’s a coffee mug on the desk (who knows what’s in it), and maybe even a vivid, shimmering city skyline in the background.

Late Show set
Although these locales are used under completely different circumstances, in both, everything in the environment, as well as the words and body language of the interviewer is deliberately arranged to extract certain types of information from an interviewee. In other words, there are many factors beyond simply what you and your participant say, that can influence what kind of information and insights the interview yields – with the environment being just one of them.

Undertaking an effective user interview requires us to retrain ourselves in how we act and speak. This doesn’t mean we can’t be friendly and casual as long as it doesn’t hinder our ability to gain useful insights. Here are some ways to ensure that you’re creating an open channel for insightful responses from your interviewees:

Build rapport – As Michael Margolis, a UX researcher at Google Ventures writes, “the data you’ll collect will suffer unless you put users at ease and earn their trust and confidence.” It’s ok to be human, but be careful when talking about yourself or relating to the interviewee, lest you introduce bias – it’s a natural tendency in humans to want to please people we like.

Don’t be the expert – Leave your ego at the door, and even play a little dumb! Try to forget what you think you know, so that you can learn something.

Any excuse to include Bruce Lee…

Any excuse to include Bruce Lee…

Make them feel like the expert – With the right encouragement, people love to show what they know, so give them an opening, and ask them to help you understand.

Control your body language – Nods, smiles, eyebrow raises, we’re all giving each other unspoken cues to indicate our emotional state, whether it’s agreement, curiosity, disgust, you name it. Much like with building rapport, non-verbal indicators like this can incline your interviewee to intuit the answers you want to hear – which aren’t always the answers you need.

Misconception 2: An interview needs to follow a set list of questions

“We have two ears and one tongue so that we would listen more and talk less.” ~Diogenes

It’s actually not all about the questions. Keep your research goals at the forefront of your mind as you craft your questions – what do you want to know, or what are you trying to prove/disprove? Use these questions as the throughline for the interview, knowing that deviating from the script is encouraged. You actually want the conversation to go in a somewhat-unexpected direction – after all, how else do you learn something new? Embrace the unexpected and dig deeper.

Your most surprising insights may come from impromptu questions spurred by your interviewee’s previous answer.

Some parts of the interview may benefit from exercises such as card sorting, or having them demonstrate how they complete a task. Remember, focus on your objectives over rigidly sticking to the script, let the conversation drift – not too far – and ask lots of follow-up questions. “Why??” is almost always a useful one, (although you’ll have to get used to sounding like a 5 year-old).

Misconception 3: People always do what they say they will do

There are two kinds of data in this world:

While face-to-face interviews are unrivalled in the richness of info they’re able to provide, it’s important to remember that people don’t always do what they say they will.

A perfect example of this phenomenon happened to us while researching how much we should charge for our blog analytics product, Filament. Upon being shown a live demo of the product, our interviewees unanimously agreed they would pay for it. Huge win, right? We also asked them how much they would likely pay for it per month – the lowest answer we got was $150/month. So we slapped a $99 price tag on it, launched & then sat back and waited for the upgrades to roll in.

…and we waited.

…and we waited some more.

They never materialized, and we eventually relented and dropped the price, because when it comes to predicting the future – even of their own behavior – your users’ estimates are terribly unreliable, especially when it comes to whether they’ll pay for things. So wherever possible, it’s a valuable use of your time to corroborate reported data with observed data (preferably collected prior to your interview).

interview workshop

Best practices for effective user interviews

Avoid leading questions

Try to avoid adding bias while delivering questions. An interview is about collecting information, not reinforcing your existing opinions. Examples of leading questions:

How easy was it to __________?

OBJECTION! Leading the witness, your honor – the question implies “__________” was easy.

Does waiting in long lines at the DMV make you frustrated?

NOPE! This question is a double-bias-whammy, in that it assumes a) the interviewee has waited in long lines at the DMV, and b) that frustration is the only thing they would feel in such a situation.

Instead, ask questions like this to get at the same insights in a more unbiased way:

How easy or difficult was it to ________?

NOICE! You’ve left the door open for a full range of answers to this question – not just the ones you were expecting.

Tell me about the last time you waited in a line at the DMV.

BOOM! Stories are some of the best data you can glean from an interviewee, as they typically contain the most detailed, and least-expected kinds of answers.

Avoid hypotheticals and guessing

In general, you should avoid the word “would.” Remember, people are really, super-bad at predicting how they themselves would respond to things in the future, especially in the context of an interview – so don’t bother asking them, not even they know for sure!

If we built this product would you use it?

How much would you pay for this?

What would you do in this situation?

Instead focus on having your interviewee describe past behaviors:

Tell me about the products you use.

How much do you currently spend on these products?

Tell me about the last time you were in this situation.

Don’t talk about your idea

If you’re using customer interviews to validate your product idea, focus the interview on validating that the problem actually exists, and that there’s a compelling need for a solution.

Stating that you have an idea or are working on something changes the tone of the interview and the participant may be more likely to positively enforce the biases you are unknowingly projecting – especially if you’re the friendly sort, and have already built up good rapport with the interviewee.

Ask the Five Why’s

The Five Why’s approach is a technique for identifying the root causes of a problem, which had its origins in Toyota’s manufacturing process, but has become popular among startups and proponents of the Lean Startup movement. It centers around asking up to five consecutive “why?” questions to get to the specific reasons behind a given user’s preference or problem. For example:

Do you use a subscription music service?

// Yes, Spotify

Why do you use Spotify?

// I like their playlists

Why do like the playlists Spotify creates?

// They have tons of them

Why do you like having lots of playlists?

// It allows me discover new music.

In this case, we see that, while it’d be easy to report that the respondent likes Spotify because of their playlists, that’s not actually an actionable insight. But by drilling deeper with more “why” questions, the interviewer was able to reveal the deeper motivation behind the participant’s preference – because Spotify allows them to discover new music. This insight can be used to drive many new initiatives, from making new music even easier to find, to changing the way they market the product.

Have them tell a story then dig deeper

Having participants tell a story instead of simply answering yes/no questions, enables you to unearth interesting topics that can be further drilled down on, such as emotions, context, outcomes, etc. It also can make the participant more comfortable because it puts the ball in their court – when they’re telling a story, they’re right where you want them: in the driver’s seat of the conversation.

What is the last movie you saw?

BZZT! The answer to this question will always be either “MOVIE NAME”, or “I can’t remember” – and you can’t really do much with either of these answers.

Tell me about last time you saw a movie

DING! This question is far more likely to be richer in details that may provide clues & hooks for insightful follow-up questions.

Good interviews lead to effective design

Well-structured user interviews can be the deciding factor between success and failure for your design, so set yourself up for a successful design and craft your questions and interviewing persona carefully – remember this isn’t just a casual chat. Effective user interviewing requires consciously rethinking how you communicate with people. Leave your ego at the door and listen.

Let your research goals drive the interview script. Allow the discussion to take its natural course and keep diving deeper to discover those hidden gems. Engage your interviewees with exercises and visuals.

There are so many best practices for user interviews, but the most important ones to remember are to have them tell you stories, and take intentional steps to avoid unduly influencing the responses you get.

Now get outside of the building, and start asking “why”!

If you’re interested in learning more about the fine art of user interviews, we’ve got all the links you can handle, right here: