Design an experimentation dashboard for Freight

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 Freight runs a logistics marketplace where merchants and shippers need reliable capacity, transparent pricing, predictable pickup and delivery, and efficient issue resolution. Product teams may run experiments across quoting, booking, carrier matching, tracking, payments, and operational workflows, but results can be hard to interpret because freight transactions are high-value, less frequent, operationally complex, and affected by market conditions.

Design an experimentation dashboard for Uber Freight that helps product, operations, data science, and business stakeholders understand whether an experiment is improving the marketplace experience without harming reliability, service quality, or unit economics. The dashboard should make it clear what changed, who was exposed, what the primary success metric is, which guardrails are being monitored, and whether the results are actionable.

Your scope is the metrics experience, not the experiment idea itself. Focus on how the dashboard should define, calculate, segment, and present experiment outcomes for Freight users such as merchants/shippers, carriers, and internal operations teams.

The experience should consider:

- Clear experiment metadata, including hypothesis, treatment/control definition, exposure unit, start/end dates, rollout percentage, and decision owner

- Primary metric definition with explicit numerator, denominator, attribution window, and why it reflects merchant or marketplace value

- Freight-specific funnel metrics such as quote request, quote acceptance, load booking, carrier assignment, pickup, delivery, cancellation, and issue resolution

- Cohort and segmentation views by shipper type, lane, region, freight type, carrier segment, load size, tenure, and operational complexity

- Guardrail metrics for reliability, on-time pickup/delivery, cancellation rate, support contacts, claims/damages, carrier experience, and gross margin

- Instrumentation quality checks, including sample size, exposure logging, event completeness, missing data, imbalance between groups, and novelty effects

- Statistical readouts that are understandable to non-technical stakeholders, including lift, confidence/credible intervals, practical significance, and duration needed

- Decision usefulness, including whether to ship, iterate, ramp further, stop, or investigate based on metric movement and operational risk

The goal is to define a dashboard that enables Uber Freight teams to make confident experiment decisions in a complex logistics marketplace, balancing merchant outcomes, carrier supply health, operational reliability, and business performance.