What Is a Funnel in Analytics?
Funnels defined: ordered steps, drop-off rates, time windows, and the classic mistakes that make funnel reports lie.
A funnel is an ordered sequence of steps toward a goal — visit → signup → activation → payment — with the percentage surviving each transition. The name describes the shape: wide at the top, narrow at the bottom, and the analysis is always about where the narrowing is steepest.
The mechanics
Each step is an observable event (pageview or custom event). A funnel report counts users (or sessions — declare the denominator) who completed step N having completed step N−1, within a time window. Window choice matters: a 1-day window on a considered purchase makes everyone look like a drop-off; 30 days flatters slow steps.
Classic funnel mistakes
- Imposing order that does not exist. Real users skip, loop, and re-enter. A strict-sequence funnel reports the loops as failures; check journey timelines before believing a step's drop-off.
- Too many steps. Eight-step funnels produce eight small numbers and no insight. Three to five transitions, each one a real commitment increase.
- Reading rates without segments. The aggregate funnel is the average of channel funnels — and channels differ wildly. Segment by first-touch source and the 'mystery' drop-off is usually one channel's tourists.
- Treating the funnel as the diagnosis. The funnel locates the leak; the why lives in the journeys — the checkout guide's read-ten-timelines method generalizes to every funnel.
Funnels worth building first
- SaaS: visit → signup → activation → paid.
- E-commerce: product view → cart → checkout → order.
- Content/lead-gen: visit → form start → submit → qualified.
Instrument the steps as events with consistent names, and the funnel is a query rather than a project — which is exactly how it should feel.