Journal · 30 Jan 2026
Exposure, trial, and habitual use are not a funnel cartoon
Funnel slides imply a single path and a moral story: people fall off because you failed them. Feature uptake is messier. Someone can be exposed, ignore the surface for a month, try it once because of a LINE message, and never return — and that can still be a correct outcome for a rarely needed job such as annual tax export.
We still use the three words because they force a denominator. We do not draw them as a triangle on page one of a readout. A triangle invites optimisation theatre: “improve step two conversion” without asking whether step two is even the job.
Habit is optional for rare jobs
If the feature is meant to be used once per year, a habit window will look cruel. In that case, define “complete use” and a quality bar (file accepted by the revenue department, not merely downloaded). Do not fake a four-day habit to look like a consumer social metric.
Exposure can shrink on purpose
Feature flags, eligibility, and empty states all reduce exposure. That can be good. A smaller, correctly eligible audience with a high trial rate is often healthier than a blast. Councils that only reward wider exposure will punish good targeting.
When students present, we ask them to say which of the three stages they are allowed to move. If engineering cannot change exposure because of a legal hold, stop proposing exposure experiments. Measure trial quality instead. The cartoon funnel never tells you which lever you actually own.