User Story:
As the Growth team, I want to conclude the current notifications experiment so that we can release the most effective versions of the notifications.
Details:
GrowthExperiments sends a notification to (most) new account holders at 48 hours. These notifications were originally developed improve engagement as part of the feature Positive_reinforcement#Levelling_up.
| Notification | 2023 audience | 2023 copy | 2025 audience | 2025 copy (T400118) |
| Get started | 0 edits | Curious about editing Wikipedia? your knowledge joins a community reaching millions. | ||
| Re-engage | — | — | 0 suggested edits, at least 1 edit and less edits than max* | Thanks for editing. Try suggested edits to develop new skills, and learn about best practices |
| Keep going | 1 suggested edit and less than max* | You’ve made X suggested edits. Keep going to learn more new skills. | 1 suggested edit and less edits than max* | |
*The maximum is defined by GELevelingUpGetStartedMaxTotalEdits for both the 2023 get started notification and the 2025 re-enage and by GELevelingUpKeepGoingNotificationThresholdsMaximum for the keep going one.
Decision:
The experiment design tested multiple variables simultaneously, which made it difficult to interpret the results precisely. However, after reviewing the data with Product Analytics, I recommend the following actions:
- Get started: move forward with the new (2025) audience and copy!
- Re-engage: move forward with the new (2025) audience and copy!
- Keep Going: revert to the previous (2023) language for the “Keep Going” notification. It seems that perhaps people appreciated the slight personalization: “You’ve made X suggested edits. Keep going to learn more new skills.”
Acceptance Criteria:
- Review superset dashboard with Product Analytics and @Sgs
- Decide on any iterative changes necessary
- Update notifications as defined above & remove any temporary code
- Release changes to all users on all Wikipedias (with GrowthExperiment enabled)
- End experiment