Funnel

How funnel analysis works

Each funnel step is a countable event — a pageview of a URL, or an explicit action recorded through event tracking, like add-to-cart or begin-checkout. The report counts how much traffic reached each step within a period and computes the transition rate between consecutive steps. Reading it is subtraction: a healthy product-to-cart rate followed by a collapse at checkout points at the checkout, not at marketing. That precision is the whole appeal — a single conversion rate says you have a problem, a funnel says where.

Funnels can be counted in two ways. Aggregate step counts compare totals at each stage over a period. Per-user path analysis instead follows identified individuals across sessions to establish that the same person completed each step in order. The second requires a persistent per-visitor identifier — which is exactly what consent rules gate. SealMetrics measures funnels the first way: each step is an anonymous aggregate count on 100% of traffic, with no individual followed through the sequence.

Why incomplete data bends the funnel

When analytics observes only consenting visitors, every funnel step is scaled down — but not uniformly. Consent acceptance varies by device, market and audience, and the missing majority does not behave like the visible minority. The result is a funnel with distorted transition rates: a checkout step can look weaker than it is simply because the segment most likely to buy is also the segment most likely to have rejected the banner or run an ad blocker. Teams then spend redesign cycles on a step that was never broken. This is the general problem of data loss in analytics applied to the report where it does the most operational damage — the one that decides what gets rebuilt.

What a funnel does not tell you

A funnel imposes a linear story on non-linear behaviour: real visitors skip steps, loop back, compare tabs and return days later, and an aggregate funnel compresses all of that into ordered totals. It cannot tell you why a step leaks — only that it does; the why needs qualitative work, testing, or session-level research tools built for that purpose. And step counts from different periods or segments are only comparable if measurement coverage was the same in both — which is precisely what consent-gated analytics cannot guarantee. See how SealMetrics reports funnels on complete data on the product page.