Generative Research
Discover what your customers actually need.
People are not very good at telling you what they want. They are excellent at showing you what they need — if you know how to watch, and what to ask.
Generative research is exploratory work: interviews, ethnography, focus groups, digital discovery. It happens early, before anything gets measured, and it answers the questions a survey cannot — why people do what they do, what problem they are actually solving, what they never thought to mention.
“Generative” here means exploratory. It is a research tradition decades older than generative AI, and it means something different. We use AI to accelerate this work — transcription, coding, pattern-finding. The interpretation is human.
What we help with
Exploratory research
A market, an audience or a problem nobody has mapped yet. Unmet needs, emerging behaviour and the why underneath a shift — before there is anything worth measuring.
Concept and product testing
Early ideas, prototypes and proposed features put in front of the people who would buy them, while changing them is still cheap. What is fatally wrong, what to sharpen, what to build.
Brand perception and positioning
What your brand actually means to people, set against the competitors they compare it with — including the associations they would never state outright.
Customer journey
The full path to purchase, from the first trigger to what happens after the sale. Where the decision is really made, and where you are losing people on the way to it.
UX and usability
Watching people use the product, site or app rather than asking them about it afterwards. Where they hesitate, where they fail, and what they never noticed was there.
Persona development
Archetypes built from depth research rather than demographics — the motivations, fears and daily routines that make a segment behave the way it does.
Creative and communications testing
Campaigns, taglines and packaging read at concept or storyboard stage, so a misreading or an unintended meaning is caught before the media spend, not after it.
Mixed methods
Qualitative and quantitative in one design: the numbers tell you what is happening, the conversations tell you why. Together they answer what neither settles alone.
What it answers
Four things a survey will not tell you.
Generative work is chosen by the kind of question, not the technique. These are the four we are asked for most.
Why they really buy
The motivations behind a purchase, rational and emotional, stated and unstated — including the ones a questionnaire cannot reach because the buyer has not yet put them into words.
What they believe before you speak
Opinions, beliefs, fears and the inhibitors that rule a brand out before it is ever seriously considered. Usually the cheapest problem to fix and the last one found.
How the category is really used
Habits of purchase and consumption observed where they happen — the home, the store, the field — including the workarounds nobody thinks to report.
What they say when you are not there
Communities, forums and open conversation read as a discovery source, for the language and the tensions a moderated room will not surface.
Methods, by name
- In-depth interviews
- One-to-one, an hour or more, following the participant rather than a script. The largest qualitative method in our ledger: 75 studies.
- Focus groups and mini-groups
- Six to eight participants, or three to four for sensitive categories, moderated so disagreement surfaces rather than gets smoothed over. 31 studies.
- Triads and paired depths
- Three participants, or two who already know each other. Enough social friction to provoke a real exchange, small enough that nobody hides in the group.
- Ethnography and in-context observation
- Watching behaviour where it happens rather than hearing it recalled afterwards — the gap between the two is usually where the finding is.
- Online bulletin boards
- An asynchronous board running days or weeks. Participants answer in their own time, come back, and react to each other — the depth of a group without the calendar.
- Auto-driving
- Participants photograph or film their own routine, then talk us through it. The account is anchored to something real instead of to memory.
- Netnography
- Reading existing online communities on their own terms, as discovery before any fieldwork is designed, and as a check on the language a questionnaire is about to use.
- Customer journey mapping
- The full sequence from trigger to purchase to repeat, assembled from interviews and observation, with the friction points named and ordered.
- Mobile ethnography
- Participants document their own environment and routine on their phone, in the moment rather than recalled afterward in an interview.
- Participant media diaries
- A running log of photos, video or voice notes kept over days or weeks — a routine as it unfolds, not as it's remembered.
- Central location tests
- A controlled venue where participants respond to a product or stimulus under supervision — the setting for taste tests, package reveals and other work that needs a controlled environment.
- Recruitment through social channels
- Targeted outreach on social platforms to reach participants a standard panel would miss — a niche user base, a specific behaviour, an audience with no existing panel.
- Global-reach online panels
- Pre-recruited online panels spanning markets beyond our own fieldwork footprint, for studies that need a broader geography than a single-market panel can offer.
Book a call with a senior researcher
Tell us the decision you're facing. A senior researcher will help you diagnose it — no sales pitch, and no obligation to run a study.
Questions we get
About generative research
When is generative research the right starting point?
When you do not yet know what to measure. If the question is “why” or “what are we missing”, exploratory work comes first — measurement without it often measures the wrong thing precisely.
Does “generative” mean you use generative AI?
No. Generative research is an exploratory tradition decades older than generative AI. We do use AI to accelerate transcription, coding and pattern-finding, but the interpretation is human.
How small can a generative study be?
Smaller than most people expect. A focused set of depth interviews often answers the decision. We size the work to the question rather than to a standard package.
What is a focus group, and when is it the wrong tool?
A moderated discussion, usually six to eight people, used when you need reactions and language quickly and when disagreement between participants is itself informative. It is the wrong tool for sensitive categories, for anything a dominant participant would set the tone on, and for individual decision journeys — those are depth interviews.
How many interviews before the findings hold?
Fewer than most people expect. Within a defined segment, new themes usually stop appearing after roughly a dozen interviews — the saturation point Guest, Bunce and Johnson documented in 2006 — and we design to that point rather than to a round number. What actually drives the total is how many distinct segments you need to read, not how large the market is.
Tell us what you need to understand.
A senior researcher will help you diagnose the decision before anyone designs a study.
