01
Exploratory (qualitative)
What is actually going on here?
Tools
Depth interviews, focus groups, diary studies, open-ended intercepts, prompted recall
Use it when
You do not yet know what you are measuring. The numbers moved and nobody in the room can explain why. Exploratory work generates the hypotheses that the other three methods then test.
Where it lies
It cannot tell you how many. A vivid quote from four women is a lead, not a market size, and teams routinely ship strategy off a single memorable interview.
With a female buyer
This is where most women's consumer research findings originate, because the deciding factors — permission, judgment, risk of being seen as frivolous — are never on a survey grid.
02
Descriptive (quantitative)
How many, how often, how much?
Tools
Surveys, panels, segmentation studies, brand trackers, purchase-frequency analysis
Use it when
You have a hypothesis and need to know its size and distribution. Descriptive work is how a qualitative insight becomes a business case.
Where it lies
It reports what people say, not why they did it. Stated-preference gaps are enormous in any category where the honest answer is socially costly — price sensitivity, beauty spend, childcare, debt.
With a female buyer
Age-and-gender segmentation is the single most common failure here. 'Women 25–54' contains at least four different buyers with opposing motivations.
03
Causal (experimental)
Does this change cause that outcome?
Tools
A/B and multivariate tests, holdout groups, geo experiments, conjoint and pricing tests, randomized message tests
Use it when
You have two credible options and need to know which one wins. Causal work is the only method that can defend a claim of cause rather than correlation.
Where it lies
It answers narrowly and briefly. A test proves a variant won this week among traffic you already had. It cannot tell you about the woman who never arrived, and it cannot explain the mechanism behind the lift.
With a female buyer
Tests run on an existing customer base bake in whoever you already convert, so a brand that alienated a segment years ago keeps proving that segment does not respond.
04
Observational (behavioral)
What did she actually do?
Tools
Clickstream and session data, purchase and returns data, shop-alongs, in-store observation, social listening, review mining
Use it when
You want behavior rather than self-report. Returns data, review text, and repeat-purchase intervals are the most under-read assets most brands already own.
Where it lies
It records outcomes and implies reasons. The record is also thinning: when an AI assistant handles comparison and verification, the deliberation disappears from the log and the shortened path gets misread as a simpler decision.
With a female buyer
Consultation is invisible in behavioral data. The group chat, the partner, the mother, and now the chatbot all sit between add-to-cart and purchase, and none of them show up in your analytics.