March 2026 · Qualitative Research
The focus group is lying to you.
Not maliciously. Respondents perform — for the room, for the moderator, for themselves. Good research designs around the performance.
March 2026 · Qualitative Research
Not maliciously. Respondents perform — for the room, for the moderator, for themselves. Good research designs around the performance.
Put eight strangers in a room with a two-way mirror, hand them snacks, and ask them what they think about your brand, and you will learn something with great reliability: what eight strangers say about your brand in front of seven other strangers.
Whether that's what they believe is a different question. Whether it predicts what they'll buy is a third question, and the gap between those three things has sunk more strategies than any competitor ever has.
I want to be careful here, because the cynical version of this take, "consumers lie, ignore them", is wrong and gives people license to skip research entirely. Respondents aren't lying. They're doing what humans do in groups: managing how they're seen. The guy who says he buys the premium option "because quality matters" isn't deceiving you. He's giving you the version of himself he'd like to be true, in front of an audience, out loud, for the first time.
The performance has patterns. People overreport aspiration and underreport habit. They claim price matters less than it does, because admitting you buy on price feels like admitting defeat. They converge: the first confident opinion in the room bends every opinion after it. And they become instant experts: ask someone to critique a logo and they will critique the logo, with conviction, whether or not they'd ever have noticed it in the wild.
The fix isn't abandoning qual. It's designing studies where the interesting data isn't the testimony. When I ran consumer research for outdoor and resort brands at Nemo Design, the sessions that changed strategy were almost never the ones where we asked people what they thought. They were the ones where we watched people do something and asked questions only after the behavior had already told on them.
Show someone three jackets and ask which is best, and you get a lecture on waterproofing. Ask them to pack for a trip they're actually taking, and watch which jacket goes in the bag. The bag doesn't perform. Diaries beat panels for the same reason: a week of "what did you actually do Tuesday" produces less flattering, more useful data than an hour of "what do you usually do."
One-on-ones beat groups for anything with identity stakes. And within any interview, the throwaway moments outrank the answers: the pause before "sure, I'd pay that," the apology before an honest opinion, the moment someone checks the moderator's face before finishing a sentence. That's where the real answer lives.
Every question teaches the respondent what you want to hear. "How important is sustainability to you?" has a correct answer, and everyone in the room knows it. The craft is asking questions that don't have a virtuous option: "Walk me through the last one you bought. Where were you? What almost stopped you?" Specific, past-tense, concrete. Memory is harder to perform than opinion.
If a research report tells you what customers said, treat it as an opening bid. Before it drives strategy, ask three things. Did we observe any behavior, or only collect statements? Were the questions past-tense and specific, or hypothetical and flattering? And would a respondent have had to embarrass themselves slightly to tell us the truth, because if the honest answer was also the comfortable answer, you haven't learned anything yet.
The focus group isn't lying to you. It's telling you exactly what it's designed to tell you. Design it to tell you something else.