What I got wrong in my first five user interviews

I read the books first. Portigal, twice. And then, in my first real interview, my second question was — verbatim — “so would a weekly summary email be useful for you?” The participant, a polite human being, said yes. So did the next four.

That “yes” was worth nothing, and it took me until interview six to understand why viscerally, rather than theoretically: people answer hypothetical questions with their self-image, not their behavior. The participant wasn’t lying. She really believed she’d read a weekly summary. She also had 3,000 unread emails, which I only learned when I finally asked her to show me how she handles email instead of asking what she thinks about it.

Three specific mistakes, so I stop repeating them:

I asked about the future instead of the past. Every question that starts with “would you” now gets rewritten to “tell me about the last time”. The past has details, friction and workarounds in it. The future only has optimism.

I rescued every silence. My recordings are painful to listen to: the participant pauses to think, and within one second I’m offering them a multiple-choice menu of possible answers. The one time my mic glitched and I stayed quiet, the participant filled the gap with the single best insight of the study.

I ran interviews to confirm, not to learn. I had a solution in my head, and my guide was a sales funnel for it. The fix wasn’t a better guide — it was writing the research questions with the team before touching interview questions, and having a colleague check each prompt for smuggled hypotheses.

The lesson file for all of this is in the UX Research track — this post is just the scar tissue.