For Growth
Measure a growth experiment by the steps customers take.
Engineering ships the code that records your experiment, and that code keeps changing between releases. Skene names the tracking the experiment is missing before it launches, then compares the result with your target when the measurement window closes.
Growth Experiment · Team invitations
38.6% activation
Target not reached
Product preview
Growth Experiment · Team invitations
Start with the outcome Growth wants to change.
- Growth’s goal
- More invited teammates become active
- Audience
- Existing customers
- Target and window
- 40% activation in the first 14 days
Before launch / Measurement requirements
Skene shows which step has no event yet.
Two milestones define the experiment.
Skene maps the events that make the growth experiment measurable, and in this earlier state the activation event is missing. That is a tracking gap, not zero activations.
invitation_sentFound in code
teammate_activatedRequired event missing
Engineering gets instructions for the missing event.
Skene supplies the requirement and the implementation instructions. A coding agent prepares the event, and engineering reviews the change before it ships.
Skene checks the required events again.
Both events are now being received. The measurement can use actual event data during the experiment’s first 14 days.
After launch / Example result
The result stays beside the target.
- Context
- Team invitations / First 14 days
- Evidence
- 1,248 invitations. 482 activations.
- Who decides
- Growth decides what to try next.
- Result state
- 38.6% activation. 40% target not reached.
Skene compares the observed result with your target
Team invitations · First 14 days
Cumulative activation rate
Activations to date divided by invitations sent to date, within the first 14 days. The line illustrates this example’s progression.
1.4 percentage points below target.
One invitation per teammate. All 482 activations occur inside the same first 14 days after launch. Available tracking is not the same as an achieved target.
The plan still needs the original event.
A renamed event can leave the experiment’s measurement incomplete. With reviews turned on, Skene checks each code change against the requirement.
Engineering reviews the correction. Skene checks it again before the result is trusted.
Use the result for the next experiment.
38.6% against a 40% target does not show the experiment caused it.
The 38.6% result falls below the 40% target, but a comparison does not prove the experiment caused the change. Growth owns that interpretation and the next version.
Connect GitHub and the source your result comes from.
Your workspace owner connects GitHub and the data source your result comes from. Skene drafts the plan and checks its evidence. Engineering reviews tracking changes.
Plan the measurement while the experiment is still a draft.
Connect GitHub or Supabase. Your first experiment is free, with no card.
Measure your first experiment free