Skip to main content
A funnel shows how people move through a sequence of steps, from a first action to a conversion.
  • Use page views and custom events as steps.
  • Conversion rate: the share of people who reached a step.
  • Drop-off: the people lost between two steps.
  • Filter funnels by campaign, source, device, country, event property and more.
Funnels count people, not events. A user who signs in on a later visit is still one user. See core concepts.

Create a funnel

Funnel is a chart type on a dashboard.
  1. Open Dashboards and open or create a dashboard.
  2. Click Add chart and choose the Funnel metric.
  3. Add two to five steps. For each step, choose Page and enter a path (for example /pricing), or choose Event and enter an event name (for example signup_completed).
  4. Choose the step order and the conversion window.
  5. Click Save chart.

Conversion window

The conversion window is the time a user has to complete all steps after the first step. The default is 24 hours. You can choose 1 hour, 24 hours, 7 days, 14 days, 30 days or 90 days. A longer window includes more completions.

Step order

Sequential (in order)

Step B must happen after step A, but other events can happen between them. This is the default. For example, in Page view /pricing → signup_started → signup_completed, a user who visits pricing, reads the docs, starts signup and then completes it counts as converted.
Lower counts than expected? People may have skipped a step, done the steps in a different order, or used a second device before they signed in. Remove an intermediate step, or use Any order.

Any order

A user must complete every step, but the order does not matter. The conversion window does not apply; steps match anywhere in the selected date range. Use it when the steps are a set of actions, not a linear journey.

Breakdown and segment

Use the controls in the chart header to:
  • Break down the funnel by Channel, Referrer, Referral, Ads, UTM Source, UTM Medium, UTM Campaign, Device, OS, Browser or Country. In the chart editor you can also enter any event property key, such as plan.
  • Choose whether a breakdown uses the user’s first touch or last touch value.
  • Limit the funnel to a saved segment instead of All users.

Read the results

Total conversion is the share of people who entered and completed the last step. Under the chart, Results and Conversion insights open one at a time. Click the open one again to close it.

Conversion insights

Open Conversion insights to see what goes with conversion. The data loads only when you open the section. It uses the same people as the funnel: the people who entered step 1, with the same steps, step filters, window, order, filters, segment and date range. Converters completed the last step. Choose a view:
  • Events: each event that people did, for example “Users with event options_open are 1.8x more likely to complete purchase”.
  • Event Properties: each property value of an event, for example “Users who did game_start where platform=steam_deck are 1.6x more likely to complete purchase”.
The insights count only what a user did before they converted: from their first event in the range until their first last-step event (and, for a sequential funnel, within the conversion window). Events of the last step itself are left out. Lift is the conversion rate of the people with the event (or value) divided by the funnel’s conversion rate. A lift of 1.5 or more shows in green, 1 to 1.5 in amber and less than 1 in red. Below each row, a bar compares the two rates and a line shows the share of people who did it. Click a row to open its table: people with and without the event, by completed and dropped off, plus the lift and the odds ratio. A row shows only when:
  • At least 10 people did it.
  • At least one of them converted.
  • Its conversion rate differs from the funnel’s rate by at least 0.2 points (funnel rate under 1%), 0.5 points (under 5%) or 1.5 points.
The section shows up to 16 rows, the strongest first, four per page. The Event Properties view looks at the 25 property values that the most people have. It skips empty and null values, objects and lists, ID-like keys (id, *_id, *_at, timestamp and similar), UUIDs, dates and large numbers. When nothing passes, it says “No significant event correlations found”. This is correlation, not cause.

Tips

  • Start with Sequential and two or three steps. Add steps when the funnel works.
  • Use the same event names everywhere. signup_completed and Signup Completed are different events.
  • Send the conversion step from your backend with server events when the browser may close before the event is sent.