Funnels

Build funnels in the Hypertune UI to:
Visualize drop-off rates between event types
Compare conversion rates across different experiment arms
Structure
A funnel is composed of a sequence of steps.
Each step can be an:
Event, e.g.
SignUpEvent
Experiment exposure, e.g.
New Editor A/B Test
orLanding Page AI Loop
For example, you can build a funnel with the following steps:
Event: PageViewEvent
Event: SignUpEvent
Event: UpgradeEvent
This lets you visualize drop-off rates between each step of your onboarding funnel.
Units
Each funnel step requires a unit ID to connect events across the user journey.
For experiment exposures, the unit ID defaults to the one used for randomization.
For event steps, you’ll need to set the unit ID manually.
To ensure accurate results, make sure the same unit ID is used consistently across all steps in the funnel.
Typically, the unit ID represents a user ID or an anonymous ID stored in a cookie. It can also be a team ID or organization ID — any identifier that’s available for every step in the funnel.
Funnel data
The first step of your funnel includes all matching events within your selected time range. The unique count shown here represents the number of unique units (users, teams, etc.) based on your configured unit ID.
Each subsequent step is filtered more narrowly:
It only includes units that appeared in the previous step.
It only counts events that occurred after the earliest matching event in that previous step.
This ensures that funnels reflect a sequential user journey rather than unrelated events.
Example:
If Step 1 includes events of type A
and Step 2 includes events of type B
:
A user logs event
A
at 10:00 AM and 11:00 AM.The same user logs event
B
at 9:30 AM, 10:30 AM, and 11:30 AM.
In this case, only the B
events at 10:30 AM and 11:30 AM are counted in Step 2 — because they occur after the first A
event for the user.
Filters
You can add filters to each step of your funnel to only select the data you're interested in.
For example, if your app runs in multiple environments it can be useful to only include data from your production environment to avoid polluting your funnel results with test data. While disabling data logging in test environments is an option, it could hinder your ability to verify functionality before a production release.

Breakdowns and segments
Each funnel step can be broken down by payload fields attached to the event or experiment exposure. Breakdowns let you segment your funnel data and compare performance across different groups.
For example, if your PageViewEvent
includes a utm_source
field, you can add it as a breakdown to see a separate segment for each utm_source
value. Breakdowns applied at one step carry through all subsequent steps, allowing you to track drop-off rates for each segment throughout the funnel.
Exposure steps can also be broken down by their dimensions, giving you a segment for each arm of an experiment.
Example:
Suppose your funnel has the following steps:
Exposure:
New Landing Page A/B Test
Event:
SignUpEvent
Event:
UpgradeEvent
You can break down the first step by the experiment dimension to compare sign-up and upgrade conversion rates for the Test and Control groups.
You can also add multiple breakdowns to a single step, or apply different breakdowns across steps. Hypertune will give you a segment for each unique combination of payload field values and experiment dimension arms.
Derived fields
High variability in certain fields can make breakdowns harder to interpret. In these cases, it’s often useful to group field values into more meaningful categories.
For example:
Group the number of basket items into ranges like
1
,2
,3–5
,6–10
, and11+
.Categorize URLs by app section, such as
homepage
,blog
, orpricing
.
You could add these grouped fields manually when sending events but this is inconvenient especially if you realize the need midway through an experiment. It also increases the amount of data sent to Hypertune.
To solve this, you can define derived fields at any funnel step. Choose the field type and set the logic to calculate its value. Once defined, derived fields behave like any other field and can be used in breakdowns and aggregations.

Aggregations
Beyond event counts, additional metrics can provide deeper insight into your experiment’s results.
For example, an increase in purchases might seem positive, but if the average purchase price drops significantly, the overall outcome may not be desirable.
You can add aggregations to any funnel step that includes numeric fields, including derived fields. Aggregations let you calculate key statistics for the selected field, such as minimum, maximum, average, and sum.

Experiment statistics
Once your funnel is set up, you can analyze performance over time and determine which variant delivers the best results.
Hypertune uses Bayesian statistics to estimate the probability that a given variant is the top performer. It also provides confidence intervals to help you validate the statistical significance of your results.
Last updated