Business Context

DiDi launched a new Post-Ride Saved Places shortcut in the rider app home screen for 50% of new users in Brazil and Mexico. The feature lets riders save Home/Work after completing a trip and then rebook in one tap on the DiDi Chuxing home feed.

Metric Scenario

In the first 14 days after launch, product leaders saw strong top-line engagement: exposed users had +11% more app opens, +8% more destination-entry clicks, and +6% more rides booked per active user versus the control group. However, finance flagged that Day-30 rider retention was flat at 28.4% vs 28.2%, and Day-60 retention was slightly down at 18.1% vs 18.6%. The GM wants to know whether the feature created real habit formation or only pulled forward trips that would have happened anyway.

You also observe that among exposed users, Week-1 repeat ride rate increased from 32% to 36%, but average rides per retained user in Weeks 5-8 fell from 3.4 to 3.1. The feature adoption rate among exposed users was 41%, with much higher adoption for commuters than occasional riders.

Requirements

1. Define the primary metric you would use to determine whether the feature improved long-term retention rather than only short-term engagement.
2. Explain how you would distinguish a true retention lift from a novelty effect or demand pull-forward.
3. Decompose the metric to identify which rider segments, use cases, or geographies drive the result.
4. Propose leading indicators that could predict long-term retention before Day-60 matures.
5. Identify guardrails or counter-metrics that could reveal the feature is increasing activity without improving user value.

Data Available

• rider_experiment_assignment
• app_event_log
• ride_orders
• saved_places_events
• rider_profile_snapshot
• crm_push_notification_log