Investigate why conversion fell after a Safety Toolkit launch

Problem Statement Description

Product context: Uber is a mobility and delivery platform; its products include rides, Uber Eats, grocery and retail delivery, freight, driver and courier tools, and marketplace pricing.

Uber has launched a new Safety Toolkit experience intended to give riders easier access to safety features such as emergency assistance, trusted contacts, trip sharing, or safety information. Shortly after launch, the team observes a drop in rider conversion, especially among business travelers, and needs to understand whether the decline is caused by the new experience, measurement issues, rollout conditions, or unrelated marketplace factors.

Your task is to investigate the conversion drop as a product RCA. Assume the affected workflow includes riders opening Uber, entering pickup and destination, reviewing available ride options, and confirming a trip. Business travelers may have additional constraints such as time sensitivity, expensed rides, airport or hotel pickup patterns, repeat routes, corporate profiles, and lower tolerance for friction during booking.

You should frame how you would diagnose the issue, what data you would inspect, how you would segment the problem, and how you would determine whether the Safety Toolkit launch is responsible. Focus on the investigation approach rather than jumping to a fix.

The investigation should consider:

- The exact conversion metric affected, including numerator, denominator, funnel step, time window, and whether the drop is statistically and practically meaningful.

- Instrumentation checks to confirm whether tracking changed during the Safety Toolkit launch, including event firing, attribution, platform versioning, and experiment exposure.

- Funnel segmentation by business traveler status, geography, platform, app version, trip type, airport versus non-airport rides, corporate profile usage, and new versus repeat riders.

- Rollout analysis, including treatment versus control, launch timing, feature exposure, localization, device compatibility, and whether the issue appears only in specific markets.

- User-behavior hypotheses, such as added friction, confusing safety messaging, increased perceived risk, modal interruptions, slower load times, or conflicts with the booking flow.

- Marketplace and external factors, such as price changes, ETA increases, driver availability, local events, competitor activity, or policy changes affecting business travel demand.

- Evidence needed to prioritize hypotheses, including quantitative funnel data, session replays or UX telemetry where appropriate, customer support themes, app performance logs, and rider feedback.

- Mitigation and prevention paths, including when to pause, roll back, ramp down, or continue the rollout while improving monitoring and launch readiness.

The goal is to demonstrate a structured RCA that protects rider safety while isolating the real cause of the conversion decline, quantifying business impact, and identifying the right decision points for product, engineering, data, and operations teams.