How user flow analysis works
Starting from an event (or working backward into one), the analysis follows what users did next, and next after that, aggregating the most common paths. The result is usually a Sankey diagram: branching bands whose width reflects how many users took each path. Where a funnel tests a sequence you chose, a flow reveals the sequences users actually chose.
Why it matters
Flows are exploratory. They surface the routes you didn’t plan for — a feature people reach by an unexpected path, a loop where users get stuck, or a common exit point right before a key action. That’s often where the next funnel or experiment comes from.
Forward and backward
Looking forward from an event answers “where do people go after this?” Looking backward into an event answers “how do people get here?” Both are useful — the backward view is especially good for understanding the paths that lead to conversion or to churn.
A worked example
You expect new users to go signup → create_project. A forward flow from signup
instead shows the busiest path is signup → view_pricing → exit: a sizable
branch detours to pricing and leaves, never reaching the product. No funnel would have caught it, because you’d never
have drawn that step. The flow surfaced the question — now you build a funnel to size it and a change to test.
How Pug does user flows
Pug’s User flow insight renders a Sankey of the paths users actually take between events — where they go next and where they drop. Combined with autocapture, you get rich paths without instrumenting every interaction, and a complementary funnel once you know which sequence to measure.