How do I analyze user paths?

Short answer

Open All-platform analytics › Behavior › Behavior analysis and choose the Paths model. Pick what a step is (page or screen, event, or both), the direction (forward from a start, or backward from an end), the number of steps (2 to 10, default 5) and optionally a start or end anchor. The chart shows the most common next (or previous) steps, with drop-off at each step on forward paths.

Path analysis needs no steps defined in advance. Use it to discover what people actually do; use a funnel to measure completion through steps you already know.

Set up a path

1. Node type. Page or screen, event, or both. On the web a page is the view path; on apps and mini programs the screen name is used. Event mode leaves out internal events whose names start with a double underscore.
2. Direction. Forward from a start shows where people go next (the default). Backward from an end shows how people got there, for example the steps before purchase.
3. Steps: 2 to 10, default 5.
4. Nodes per step: 3 to 15, default 8. Less frequent nodes are folded into Other.
5. Anchors. Set a start anchor, an end anchor, or neither. With neither, forward paths start at the first step of each session and backward paths at the last.
6. Counting: sessions (default) or people. Switch to people for journeys that span several visits.
7. Repeated nodes collapse by default, so reloading the same page counts once.
8. Exclusions (up to 20 patterns, wildcards allowed) remove noise such as /favicon or heartbeat events.
9. Clean-up rules (up to 10) rewrite matching paths into one label, for example turning /users/123 and /users/456 into /users/:id.

Read the result

The chart is a flow from one step to the next. The number on a node is how often that node was passed at that step. On forward paths, each step also shows drop-off: what flowed in minus what continued. The summary at the top gives starting volume, node count, step count and total drop-off.

How to use it:

• Start from a page that matters (pricing, sign-up, a new feature) and look for the biggest unexpected branches.
• Go backward from a goal to see which pages or events usually come before a conversion.
• Compare with your intended flow: large flows back to an earlier page often mean people are missing information.
• Export gives source node, target node and count for further analysis.

Limits

• Up to 50,000 transitions are kept before the per-step trim, so very rare paths are dropped first.
• Drop-off is only meaningful on forward paths.
• An anchor uses the earliest start or the latest end within the session (or person). Without an anchor, a session contributes only its first or last few steps.
• There is no breakdown and no comparison period on paths.

FAQ

Why do I see "Other" everywhere? Each step keeps only the most frequent nodes (8 by default). Increase nodes per step, add clean-up rules for URLs with IDs, or exclude noisy pages.

Path, funnel or goal? Paths discover routes, funnels measure conversion through fixed steps, goals count one outcome. See Funnel vs goal vs path.

Why are counts lower than my pageviews? The default count is sessions, not pageviews, and repeated identical nodes collapse.

Related

• Build a funnel
• Read funnel drop-off
• User flow: designed vs actual paths