Business Context

Uber’s core ride-hailing product serves ~25M daily trips globally. Customer Satisfaction (CSAT) is a top-level KPI reviewed weekly by the GM because it predicts rider retention and complaint volume, and it influences marketplace health (driver supply) through support load and refunds. Over the last 6 weeks, CSAT for completed trips dropped from 4.72 to 4.58 (-0.14) in the US and Canada, while trips are flat (+1%) and on-time pickup rate is only slightly down (-0.6pp).

A cross-functional incident review is underway. Product suspects a recent pickup-pin UX change increased confusion. Operations suspects airport queues and driver cancellations. Support suspects a policy change reduced goodwill credits, increasing negative sentiment. You are the analytics lead asked to analyze survey results and recommend improvement efforts that can be executed in the next quarter.

Metric Scenario

After each completed trip, ~8% of riders receive a CSAT survey within 30 minutes. The survey includes:

• Overall rating: 1–5 stars
• Multi-select reasons (e.g., “Driver behavior”, “Pickup issues”, “Price”, “Vehicle condition”, “App issues”)
• Optional free-text comment

You also have event logs for the trip funnel (request → match → pickup → dropoff), pricing, ETA accuracy, cancellations, and support contacts. Stakeholders want a clear diagnosis: Is the CSAT drop real, where is it coming from, and what should we do first?

Data Available

What You Need To Produce

1. Define the CSAT KPI precisely (who is included, time window, how to handle missing responses, and whether to weight by trips or by respondents).
2. Validate whether the CSAT decline is statistically and operationally meaningful (e.g., response-rate shifts, sampling bias, seasonality, city mix changes).
3. Decompose the CSAT change into actionable drivers using at least two complementary approaches:
   • A metric decomposition (e.g., mix vs within-segment change)
   • A driver analysis (e.g., top reasons, correlation/attribution to operational metrics)
4. Prioritize 3–5 improvement efforts with estimated impact, effort, and risk; include how you would measure success and what guardrails you’d monitor.
5. Recommend a weekly CSAT dashboard: what leading indicators should be tracked to catch future CSAT regressions earlier.

Constraints

• You have 5 business days to deliver a readout to the GM.
• Any proposed changes must not increase rider cancellations by >0.3pp or refunds/credits by >5%.
• The pickup-pin UX change can be rolled back for a subset of users, but only via an A/B holdout in the next release cycle.