But what if it’s not statistically significant?

I met with some company stakeholders at a client’s office the other day, explaining our process and how we’d be using interviews and qualitative analysis to do a deep dive into the minds of their customers. Heads nodded, notes were taken, and one man seemed fit to burst into applause.

All sailing was smooth until I said, “We’re going to interview twenty of your customers to get this data.”

“Wait a second,” said a guy at the other end of the table who’s an avid reader of Devil’s Advocate Quarterly, “How will twenty interviews be statistically significant?”

It’s a valid question, and having a close business relationship with The Devil himself, one that I pondered when I first was introduced to ethnographic methods. Because the possibility is real: if we only survey twenty people, what if they’re not representative of the whole customer base?

The answer lies in that word back there, “survey.” Doing an in-depth customer interview is about as far from a survey as you can get, and collects data at a depth that written surveys aren’t capable of. Instead of just collecting standalone data points — which can then only really be correlated with outcomes, rarely directly linked — these in-depth interviews get to the root cause of consumer behaviors, making it far easier to build out a rich, understandable story about a consumer’s true experience.

I like to think of it this way: imagine taking your car to the mechanic because it’s doing that weird thump thump sound again. You show up, and the friendly mechanic hands you a survey. “Here, fill this out,” she says (you thought the mechanic was a man, didn’t you. Surprise.) You input a bunch of data about yourself —how often you drive, how much money you make, how satisfied you are with your driving experience on a scale of 1-5. The mechanic correlates your answers with her database… “According to our data, most people made the problem go away by getting a new car.”

That’s not how it really works, of course — your mechanic is able to actually speak to your experience, hear your story about driving over the curb, and actually observe your car for a root cause. You’ve got a flat, and she didn’t have to interview a statistically significant group to uncover that.

That’s what we’re looking for with these customer interviews — stories that get to the underlying cause of why certain outcomes are happening, not just correlated inputs and outputs. Understanding the root cause makes it far easier to design services and products that fill a real need.

Quick caveat: you’ll know this process is working when, by the 8th interview, you can pretty much guess what the person is going to say regarding their general needs, desires, and struggles. If you’re hearing lots of different stories, though, it may mean there are more divergent customer journeys than expected. In these cases, yes — you’re going to have to do more interviews until you can bucket out these groups and start to see predictable patterns.

Happy interviewing.