What users actually click, use, and abandon — not what they say or recall. The quantitative ground truth of behaviour, and the starting point for knowing which questions to ask.
▸ Try the interactive toolBehavioural Analytics is the practice of analysing data about what users actually do in a product — what they click, which features they use, when they drop off, how long they engage, and what action sequences precede retention or churn. Unlike survey data (what users say) or interview data (what users recall), it's a record of real behaviour.
Its strength is also its limit: analytics tells you precisely what happens and how much, but never why. A drop-off chart shows exactly where users leave and in what numbers — and is completely silent on the reason. So behavioural analytics is best understood as the instrument that tells you where to point your qualitative research: the numbers find the anomaly, and interviews or contextual inquiry explain it.
Can: what users do, how often, where they drop off, what sequences precede an outcome — at scale, objectively.
Can't: why they do it. For that, the numbers point you to the right qualitative question.
Behaviour and self-report diverge constantly. Analytics is the antidote to building on what users claim — but mistaking the “what” for the “why” is its classic trap.
| The shortcut | What it costs | What it gives you instead |
|---|---|---|
| Trusting self-report | What users say they do differs from what they do. | Analytics records actual behaviour, objectively and at scale. |
| Guessing where problems are | Teams debate where users struggle without data. | Drop-off and funnel data pinpoint exactly where, and how much. |
| Anecdote-driven decisions | One loud complaint drives a roadmap. | Analytics shows whether the anecdote is widespread or rare. |
| Mistaking what for why | Reading a cause into a number that only shows behaviour. | Analytics locates the question; qualitative research answers it. |
Track the key actions, funnels, and sequences tied to your outcome — not everything. Un-instrumented behaviour is a blind spot; over-instrumented data is noise.
Look for the surprising drop-offs, the unexpected sequences, the features used far more or less than expected. The anomaly is where the insight — and the next research question — lives.
When you see a sharp drop-off, the temptation is to assume the reason. Note the what precisely and hold the why as an open question, not a conclusion.
Take the specific behavioural finding to interviews, usability tests, or contextual inquiry. “Users drop off here — let's watch some do it” is far sharper than open-ended research.
Combine the behavioural what with the qualitative why into a complete picture, then act — and re-measure behaviour to confirm the fix worked.
Analytics showed a sharp, consistent drop-off at one step of a flow — precise and unambiguous about where and how many. The team immediately assumed they knew why (“the form's too long”) and shortened it. The drop-off barely moved.
Only when they pointed usability testing at that exact step did the real cause appear: users weren't deterred by length — they hit a moment of confusion about what a field meant and abandoned rather than guess. The analytics had perfectly located the problem and said nothing about its cause; the assumed why was wrong, and the data couldn't have corrected it.
The deliverable is the precise behavioural pattern (what, where, how much) paired with the open why-question it raises for qualitative follow-up.
| Analytics gives you | It does NOT give you |
|---|---|
| Where users drop off | Why they drop off |
| How many, how often | What they were feeling |
| What sequences precede churn | The cause of the churn |
| The anomaly to investigate | The explanation — that needs research |
Behavioural analytics is the most reliable answer to “what is happening?” and no answer at all to “why?” Its highest use is as a targeting system — it tells your qualitative research exactly where to aim.
The discipline that separates good analysts from dangerous ones is refusing to invent the why. A drop-off chart is almost irresistibly suggestive — the moment you see it, a plausible explanation forms, and it feels like knowledge. But the number contains no causal information whatsoever; the plausible story is a guess wearing a chart's authority. The teams that win with analytics treat every striking number as a precisely-located question, then spend a little qualitative research answering it — rather than acting on the confident, untested story the data seemed to tell.
The cardinal sin. A number shows behaviour, not cause. Investigate, don't assume.
Tracking it all creates noise, not insight. Instrument what ties to the outcome.
Analytics alone is half the picture. Pair the what with a researched why.
Big, flattering numbers that don't tie to behaviour or outcomes waste attention.
Analytics finds the anomaly; interviews, usability tests, and contextual inquiry (Tools 05, 06, 17) explain it.
Cohort analysis (Tool 11) adds the time and group dimension to raw behavioural data.
When analytics suggests a change, A/B testing (Tool 12) proves whether it actually causes improvement.
Behavioural analytics is the foundation of the funnel, retention, and engagement metrics in Module 5.
Imagine (or find) a drop-off or usage number that surprises you. Write down the immediate explanation that springs to mind.
Now treat that explanation as a hypothesis, not a fact: what qualitative research — which users, doing what — would actually confirm or kill it?
The reflexive explanation you wrote first is exactly the kind of invented “why” that analytics can't support — noticing the urge to believe it is the whole skill.