Discovery Phase Made Easier: The Joy of Self-Interviews

How do people really feel? What drives them? What are their most important dreams and problems, and how do they deal with them today? “Discovery” questions like these are the foundation for creating any good customer experience, but it’s often very hard to get them answered. Live one-on-one interviews are the gold standard for collecting that information, but they’re often so time-consuming, difficult to set up, and expensive that we don’t do them as often and as thoroughly as we should.

I’ve been experimenting with an alternate approach, which we call “self-interviews.” It’s a way to do discovery that’s much faster and more convenient than live interviews, but that can still produce high quality insights. It’s easy to do even if you’re not trained in research, and it deserves to be in your insight-gathering toolkit. In this article I’ll describe the technique, the things you can learn from it, and the steps you can take to start running your own self-interviews.

What is a self-interview? 

The idea is simple: Instead of scheduling a live interview with someone where you ask them questions in real time, you write out your questions ahead of time, send them to the customer, and on their own they record a video in which they answer the questions. Using a well-designed human insight system, it’s easy to set this up – it’s just a standard usability test, but instead of asking someone to use a website or play with a prototype, you have them read questions from the screen and respond to them.

The self-interview has a lot of advantages:

• It’s very easy to recruit and schedule, since that is done automatically by your human insight system
• You get a faster turnaround time than live interviews, since all of the sessions can be recorded simultaneously
• It consumes less of your time, because you don’t have to be on the line with any of the participants when they’re taking the test
• You can save even more time by using the analysis capabilities in your human insight system to rapidly parse the results. For example, in the UserTesting system, I can play back videos at high speed, read the transcript, and use emotion analysis to jump to the most important answers. You don’t have to watch the complete videos.
• No-shows are a thing of the past, since you’re only analyzing videos that actually get completed

Using this process, I can usually compress a discovery process that would have taken days or weeks to a single day.

There are also some disadvantages to self-interviews:

• Because you’re not talking with the customer live, you can’t ask immediate follow-up questions (although some human insight systems will let you recontact participants later)
• If the customer is confused by a question, you don’t have the opportunity to explain and correct it.

When those problems happen in a self-interview, the fix is to revise the test and run it again. But that does chew up a lot of the time you were hoping to save.

When I tell people about the self-interview process, they often ask me whether the quality of responses in a self-interview is as good as you’d get from a live interview. As far as I know, there haven’t been any controlled experiments on this, but in my experience the responses are very good. In some cases I believe they may be even better than you would get from a live interview.

I think people behave differently when talking to their computer or smartphone than they do when talking to a live human being. Self-interviews are sometimes almost confessional in their intensity and candor. You’ll also see people in their natural settings, sometimes in pajamas or a messy room. In live interviews, I think they’re much more self-controlled and aware of putting on a good appearance. The thing I can tell you for sure is that the responses are different from what you get in a live interview, and so it’s useful to experiment with this approach to learn what it can and can’t do for you.

In my own work, I’ve used self-interviews to probe a number of very delicate issues, including feelings about the Covid pandemic, reactions of Black people to the Black Lives Matter protests, and thoughts on the 2016 and 2020 Presidential elections.

Here’s a video of people answering questions about Covid. These participants are all talking to themselves, not an interviewer:

The Black Lives Matter and election coverage were both written up on our sister site, the Human Empathy project. You can find those reports, and results videos, here:

• Black Lives Matter
• Undecided voters and the 2020 US Presidential election

How to set up a self-interview

If you’re ready to try your own self-interviews, just follow these steps:

Make sure your insights system can record faces. Unlike a usability test (where it is helpful but not essential to see the participant), you must be able to see the participants throughout a self-interview. We don’t recommend using a system that can only record video in short clips or responses to selected questions. Often the sighs and tone of voice while people read a question will tell you as much as their actual answers.

Configure it as a web test. Most human insight systems do not yet have a separate setting for a self-interview. Instead you set your test up as a website usability test, but don’t take people to a website. Some systems let you leave the browser field blank, so a website is not displayed on screen during the test. If you have to put in a URL, you can set it to a blank one like http://www.blankwebsite.com/.

Be completely up front about what you’re doing. Since this is an unusual type of test, you need to give people an idea of what to expect. A good human insight system will let you input some text describing what your test is all about. You should tell participants the nature and subject of the test, and if you’re asking issues that might make people uncomfortable, you should be as clear as possible about what you’re doing with the video, and who is sponsoring the test. This is a departure from usual test practice. Normally you want to keep your company name confidential, but if you want candid answers, you need to reassure them that you don’t have a hidden agenda. For our Covid self-interviews, we used this introduction:

This is not a usability test. UserTesting is doing research on the Covid-19 pandemic and we want to learn more about your reactions to it. We’re going to show you a series of questions about the pandemic. Please spend about a minute answering each one.

Ask a series of verbal questions. A good human insight system will let you ask what’s called “verbal” questions. Those are questions that the participant is prompted to read and answer out loud, which is exactly what you want for a self-interview. 

Use the system’s automated analysis tools to save time. One advantage of having the participants record their answers is that you don’t have to watch every minute of every video. A good human insight system will let you play back the videos at high speed (I recommend 1.5x or 1.75x) and read the responses in a transcript (most of us can read far faster than we can talk or listen). Some systems have AI-driven sentiment analysis, which flags key points in the videos for you, so you can jump straight to them. Once you get used to how this works, you can save an enormous amount of time. You can use that time to shorten the discovery process, or to dig deeper and do more interviews. 

Rating scale questions also work extremely well in a self-interview. You give participants a statement and ask them to agree or disagree on a numerical scale. For example:

Please read the following statement, rate how strongly you agree or disagree with it, and then explain your answer out loud:

“I am afraid to ride on public transit due to the Covid-19 pandemic”

1. Strongly disagree   2.   3.   4.   5.   6.   7. Strongly agree

Start them off slow. People love talking about themselves, so it’s often very effective to ask them a couple of general background questions before you turn to your specific subject. That can get them in the mood to talk. For example, “Describe your job. What are you responsible for, and what are the biggest barriers to getting them done?”

Here are a couple of other questions that work very well in a self-interview:

• “If you had a magic wand, what’s the one change you would make to (insert issue here), and why?”
• As the end of the test, ask “Is there anything else that we should have asked you, and why?” Often participants want to be helpful, and they’ll volunteer information that you would never have thought to ask about.

Always remind them to explain their answers, and tell them how long you want them to talk. Some people become quiet when they’re reading a screen, so it doesn’t hurt to remind them frequently to speak up. When you want a thorough answer, it’s a good idea to also give them a time prompt, something like “Please spend at least two minutes describing…”

Follow these pointers and soon you’ll be doing discovery all the time. 

Let us know what you think! Please share your questions and experiences in the comments below.

The opinions expressed in this publication are those of the authors. They do not necessarily reflect the opinions or views of UserTesting or its affiliates.