The answer was already in the room
In 2010 we ran focus groups about camcorders. Participants ranked image quality, memory and zoom, exactly as the brief asked. The sentence that ended the category was also said out loud that night — as an aside, while talking about something else.
The study was straightforward. A manufacturer wanted to know how Brazilians chose a camcorder: which attributes mattered, which concepts appealed, what people would pay. We recruited across ages and social classes, moderated four groups in São Paulo, and delivered what was asked for.
The groups worked. Participants were articulate about what a camcorder was for, and they were unanimous about it. The first words out of their mouths, before any prompt about features, were about time.
“It's to remember something unique. Something you'll never see the same way again.”
Woman, 30s, São Paulo · translated from Portuguese
Then the discussion moved to attributes and did what discussions about attributes always do. Image quality. Memory — how many hours it holds. Zoom. Size. Whether the video would upload straight to the social network everyone in the room was using at the time. Participants ranked them, scored them, argued about them. Good, usable material.
The aside
Halfway through, the moderator asked something incidental: whether they carried a camera when they went out. The answers were casual, almost throwaway, and they were about logistics — bag size, fear of theft, convenience.
“My camera is a bit big, so the phone does the job.”
Woman, 20s, São Paulo · translated from Portuguese
And, from another participant, the full logic laid out in a single breath: if I know I'm going out to take pictures, I bring the camera; otherwise I just bring the phone. If there's any risk of losing it or being robbed, the phone. A device you already carry beats a device you have to decide to carry.
Nobody in the room, including us, treated this as the headline. It wasn't an answer to the question being asked. It was a small practical remark about handbags.
Within a few years the consumer camcorder was effectively gone, and the compact camera followed it. Not because phones filmed better — in 2010 they very obviously did not — but because the phone was already in the bag.
People are reliable about needs and unreliable about objects
This is the part worth keeping. The participants were not wrong. Read the transcript again and they are strikingly accurate about the job: hold on to something that will not happen twice, keep the way a person moved and spoke, be able to go back. Every one of those sentences is still true in 2026. All of them are now satisfied by a phone.
What they could not do was predict the object. Asked to imagine a better camcorder, they described a slightly better camcorder — smaller, more memory, easier to upload. Asked how they actually behaved, they described the future without noticing.
That gap is not a flaw in respondents. It is a flaw in questions. A brief that asks “which attributes drive choice in this category” cannot return an answer that says “this category is about to stop existing.” The material was there; the question wasn't.
This is a re-reading of one study, sixteen years later, and hindsight makes patterns look obvious that were not. Four groups in a single city cannot describe a national market. We are also not claiming we spotted this at the time — we did not, and that is precisely the point. The claim here is narrow: the signal existed in the transcript before it existed in the market.
What we do differently because of it
Three habits, all cheap:
Code the asides. When a participant answers a question you did not ask, that fragment gets tagged and kept, not trimmed. Substitution behaviour almost always shows up as an aside, because from the participant's point of view it isn't news — it's just what they do.
Ask about the last time, not about the category. “What do you look for in a camcorder” produces an inventory of attributes. “Tell me about the last thing you filmed, and what you filmed it with” produces the truth, including the devices the brief never mentioned.
Separate the job from the object in the analysis. Written as two columns, the 2010 material is unambiguous: the job column is stable and the object column is already wobbling. Merged into a single ranking of attributes, it reads as a healthy category.
None of this requires new technology. It requires deciding, before fieldwork, that the interesting material might not be the material you commissioned.
We keep sixteen years of Brazilian fieldwork, and we still read the old transcripts. If you're planning a study here, we run the field.
Start a project