Consumer insights: a data point is not an insight
"62% value sustainability" is a data point. "They buy the sustainable one when it is in front of them and the price does not force them to justify it at home" is an insight: it explains a behaviour and suggests what to do. The difference matters because most briefs ask for insights and receive data, and because the good ones come out of people's language, not out of a table.
What an insight has that a number does not
A tension, an explanation and an implication. The tension is the conflict between what someone wants and what stops them. The explanation is the why, in their words. The implication is what we would do differently tomorrow if this is true.
A percentage carries none of the three, which is why the source of insight is qualitative and the number arrives afterwards to size what you already understood.
Good ones come from contradiction
When someone says price is irrelevant and then explains they wait for the sale, that is an insight. When two segments describe the same product in opposite words, that is another.
That is why disagreement inside the room is worth more than consensus, and why a report that averages six opinions usually throws away the finding. The room is built around it: participants read what the others said and push back when a different profile would, and the sentiment strip reads "split" as a legitimate result rather than as missing signal.
The part that is hard from abroad
Not the method — the grounding. An insight about a Spanish shopper that ignores what a weekly shop costs there, which chains people name, and what has been in the news that month is an insight about nobody.
Each market here carries its own currency, price habits, retail and media landmarks, register markers and a fresh context digest, so the reasoning happens inside that country rather than in a global average.
A process that fits in an afternoon
Write the decision in one sentence. Ask about past behaviour instead of future intention. Chase the contradiction with a follow-up to the rest of the room. Write down the literal vocabulary. Leave with two or three hypotheses and a validation plan. 60 answers cost €7.99, so the loop can run as many times as the question deserves.
The limit, said plainly
A synthetic archetype has never bought anything. What it produces is a well-informed hypothesis about how a segment reasons, not an observation of behaviour.
If a line from here goes into a client deck, label it: synthetic audiences, directional signal, validation plan attached. That is what gets the method used a second time.
Frequently asked questions
Context and structure. A chatbot answers in the average; here there are six profiles with age, city, income, education and register, a market context digest injected before the session, and a room where each participant hears the others. You also leave with a transcript, sentiment by segment and an exportable report, which is what turns a chat into a deliverable.
Keep reading
Last reviewed: 2026-09-14. Participants are AI-generated synthetic archetypes, not real people.