How to analyze consumer behaviour
Four ways, each with a job to do. One-off studies, segmentation analysis, behavioural tracking, and trend analysis over time. Use only the first two and you’ll get a static picture that’s already out of date. Use the third and forth well and you’ll see movement – which, in a volatile market, matters more than scale.
The textbook answer usually lists methods in parallel and calls it done. Run a survey, pull the panel data, cut by demographics, repeat. That works in stable markets. In shifting ones, it produces reports that are technically accurate and commercially useless.
A better question isn’t “which method” but “which method tells you what.” Each of the four below answers a different question. The strongest analysis uses them in sequence – shape, size, segmentation, movement – and treats the gaps between them as insight, not error.
1. One-off surveys – the shape of the market
Qualitative, quantitative, or both. Surveys tell you what people say they think, feel, want and plan to do. They’re strong on attitudes, intentions, priorities and self-reported behaviour. They’re good for getting the shape of a market quickly – who’s in it, what they say they value, which tensions sit underneath the category.
Used well, surveys set the frame for everything that follows. They surface hypotheses, map the emotional territory, and identify where the interesting commercial questions are likely to sit.
Used badly, they end up treated as ground truth. Self-reported behaviour is often the weakest form of evidence – the consumer’s story about themselves, not a record of what they actually did. A survey saying 30% of consumers plan to switch and a panel showing 4% actually switching aren’t contradicting each other. They’re measuring different things.
Use surveys to map the territory. Don’t use them as the final word on behaviour.
2. Segmentation analysis – who behaves differently, and why
Grouping consumers into meaningful sets is the step that turns description into strategy. Without segmentation, every insight is an average, and averages hide the commercial action.
The textbook default is demographic segmentation – age, income, region, household composition. It’s easy to build and easy to target. It’s also increasingly weak at predicting behaviour. Two 45-year-old women with the same household income and the same postcode can behave like completely different consumers.
The stronger approach segments by need, mindset or attitudes. ?????
Good segmentation lets you treat different groups differently. Which is the whole point – a single message to “the consumer” is a message to nobody in particular.
3. Behavioural tracking – the size of the opportunity
What people actually do – browsing, buying, switching, using – is the third layer. Panel data, loyalty data, EPOS feeds, digital behavioural tracking, app usage, subscription churn. Objective, observable, hard to fake.
Behavioural tracking is the strongest source for sizing the market and identifying pounds-and-pence opportunities. Where is the money flowing? Which categories are growing, shrinking, fragmenting? Which segments are spending, switching, drifting? Behavioural data won’t tell you why, and it tells you where very well.
The weakness is the inverse of surveys. You see what people did. You don’t see what they considered and rejected, what they meant to do, what they’d pay for that doesn’t currently exist. Behavioural data is backwards-looking by design.
Pair it with surveys and you get a 360 on the current state – what people say, what they do, and the gap between the two.
4. Trend analysis over time – the direction of travel
This is where most analysis falls down, and where the commercially interesting answers actually live.
A single cross-section tells you what’s true now. It doesn’t tell you whether that truth is forming or fading. Trend analysis over time – ideally weekly, ideally across attitudinal and behavioural layers simultaneously – is the method that catches movement. In a volatile market, movement shapes demand more than absolute scale does.
Weekly longitudinal tracking lets you do four things no other method can.
Spot the say/do gap as it opens and closes. When stated intent and behaviour diverge, something is shifting. Narrowing, and the intent is starting to convert. Widening, and mood has moved faster than action.
Test things implicitly. News events, cultural moments, inflation changes – all happen against a live backdrop. Good longitudinal data lets you watch what moved, what didn’t, and what moved in the opposite direction to expectation.
Read the direction of travel. A 30% sentiment score on the way down is a completely different commercial story to the same score on the way up. Direction is signal. Level is context.
Anticipate shifts in the shape of demand. A market growing at 3% can still hollow out in the middle or polarise at the ends. Static tracking can’t see this. Weekly tracking picks it up in the slope, not the average.
This is the method most often missing from standing analytical capability – usually because it requires continuous data collection, which has historically been expensive to run.
Putting it together
No single method wins. The strongest analysis runs them in sequence.
Start with surveys to map the emotional and attitudinal terrain. Layer behavioural tracking to size the opportunity and identify where the money is actually moving. Segment the audience using something more useful than demographics. Then run trend analysis over time to watch the shape shift, catching movement as it happens rather than describing it after the fact.
The interesting insight is usually in the gap between two of these layers. Surveys say the category is valued – behavioural data says it’s losing share. Segmentation says one group is the growth driver – trend data says the underlying mood in that group is softening. Analysts who read the gaps outperform analysts who read the methods in isolation.
And the practical output from the fourth method is the one that most changes what brands actually do. A piece of research that confirms what you already knew gets noted. A piece of analysis showing the direction is about to change gets acted on.
How Kokoro tracks this
Running all four methods in-house is expensive. Running the fourth – continuous longitudinal tracking – is the one most brands can’t sustain on their own.
Kokoro runs 2,000 consumer interviews every week – more than 100,000 a year – combined with a longitudinal community of 50 UK households tracked over six years. Attitudinal and behavioural layers in one system. Mindset segmentation built in. Trend analysis produced weekly, not quarterly.
| Traditional approach | Behavioural approach |
|---|---|
| Commission ad hoc surveys when needed | Continuous weekly tracking always on |
| Cross-sectional snapshots | Longitudinal read of direction and movement |
| Demographic segmentation | Mindset, need, life stage and behaviour |
| Attitudinal or behavioural | Both, tracked together |
| Reports what’s true this quarter | Reports what’s shifting this week |
| Findings land six weeks after the shift | Findings land in the week the shift happens |
For analysts and insight managers, the practical difference is that continuous tracking makes ongoing analysis viable without stacking up ad hoc studies every time a question comes up.
Movement matters more than scale
The question “how do you analyse consumer behaviour” usually gets answered as a list of methods. More useful: which method answers which question, and which method tells you what’s about to change.
Surveys give you the shape. Behavioural tracking gives you the size. Segmentation gives you the cuts. Trend analysis gives you the direction of travel – and in a market like this one, direction is worth more than level.
