Open vs closed funnels and funnel diagnostics

A closed funnel (the default) only counts users who entered at step 1. Someone who landed directly on the checkout page is excluded, and the completion rates read as "of those who started here". It is right for designed flows such as onboarding or checkout, and for asking how good the flow is.

An open funnel (Make open funnel toggle) lets users enter at any step. A user who skipped straight to step 3 is counted at steps 3 and 4. The numbers at later steps rise, often sharply, and the funnel now answers "how many people reached this point, however they got here". Use it for loosely ordered journeys such as content paths, or when deep links and ads land people mid-flow.

When a step leaks, work through the diagnostics in order:

• Breakdown by Device category or Browser: a leak isolated to one platform is a bug, not a design problem.
• Next action: if users who abandoned checkout next fired view_cart or page_view of the shipping page, the price or a form is the issue; if they fired nothing, they left the site.
• Elapsed time: a step that takes users minutes instead of seconds is a confusing step.
• Trended funnel: did the drop appear on a specific date? Line that date up with deploys and campaign launches.
• Segments: compare new vs returning, or a segment of a campaign's users, as separate funnels.

Two accuracy notes. A funnel spanning sessions works because funnels are user-based, but a step condition on a session-scoped dimension (such as Session source) is evaluated per event's session, which can exclude users who switched sources. And funnel user counts respect the reporting identity: with User-ID, a user who added to cart on mobile and bought on desktop completes.

All funnel data is subject to the retention window and sampling like any exploration.