Thin-slicing user interviews

One of the core ideas in Malcolm Gladwell's new book Blink is thin-slicing, our ability to make accurate judgements about people or events in a very short time frame.

Early in the book Gladwell describes an experiment where strangers made surprisingly accurate assessments of students' personalities based only on looking at their dorm rooms for 15 minutes. On three of five measures used in the experiment--conscientiousness, emotional stability and openness to new experiences--the strangers made better assessments than the students' friends.

This got me thinking about my appoach to user interviews, which normally involves a getting-to-know-you interview to understand how a person thinks about a problem, application, product or whatever.

A typical project might involve 20 hours of user interviews, for which I usually estimate about an hour each. I like to keep most of the interviews between 20 and 30 minutes--I find people often don't have a lot to say after that. Of course there are some people that have almost nothing to tell you (either because they don't care or they can't articulate their real needs). And then there are people who are a goldmine of information and insight about your project or client. So 20 hours of interviews might break down like this:

But after reading Blink I've been thinking of breaking it down like this:

Same amount of time total, but I'm doing two things differently: 1) I'm getting a much broader sense of the high-level user needs (I think this helps inform my judgement), and 2) I'm devoting more of my time to the people who give good interviews.

Doing this well would require access to a large group of readily-available users (I'm assuming there's no increase in recruiting time), and a script to help discover both high-level needs and good second-interview prospects. I'd also want the script to focus more on eliciting reactions than just gathering information. (In another experiment discussed in Blink it was shown that couples who displayed subtle signs of contempt during a short conversation almost always got divorced. I guess we'd want to know if someone felt contempt for our application, intranet, whatever.)

There's a more practical side to this idea, too. We've been doing a lot of health care work lately, and it's often hard to get 30 minutes with a doctor. But five minutes with a doctor--in his or her work environment--is much easier to get, and it might be worth more than 30 minutes on the phone. That little difference adds up in my thin-slicing scenario, especially if you also get six good, in-depth interviews.