# Design an End-to-End Customer Delivery Experience Dashboard Design a report or dashboard that follows a customer order from placement through delivery and shows whether the experience was good or bad. The audience needs both an executive health view and enough diagnostic depth to identify where the journey failed. ### Constraints & Assumptions - Events can be late, missing, duplicated, or timestamped by services in different time zones. - A cancelled order does not reach every journey stage. - Customer sentiment, promised time, actual time, defects, and support outcomes can disagree. - The dashboard must support trends without exposing individual customer data broadly. ### Clarifying Questions to Ask - Who is the primary user and which decisions should the dashboard drive? - What defines an eligible order and when is its outcome final? - Is good or bad a composite label, a hierarchy of failures, or separate dimensions? ### Part 1 — Journey model Define stages, timestamps, and data-quality rules from order placement to final outcome. #### What This Part Should Cover - A canonical order grain and event contract - Stage-specific denominators and terminal states - Late-data and duplicate handling ### Part 2 — Metrics and layout Choose headline, funnel, timing, quality, and support metrics and arrange them for overview-to-diagnosis navigation. #### What This Part Should Cover - Promise accuracy and tail latency - Cancellation, defect, refund, support, and satisfaction measures - Cohort filters and contribution analysis ### Part 3 — Validation Explain how to prove the dashboard reflects customer experience and remains trustworthy after launch. #### What This Part Should Cover - Reconciliation to source systems - Label validation against qualitative evidence - Freshness, completeness, and anomaly monitoring ### What a Strong Answer Covers - Metric definitions with denominators - A diagnostic information hierarchy - Data quality and privacy controls ```hint Avoid one opaque score Use a small set of interpretable outcome dimensions and a clear failure hierarchy. A composite may summarize, but users must be able to see which stage caused it. ``` ### Follow-up Questions - How would you distinguish merchant delay from courier delay? - What alert would indicate that the dashboard itself is wrong?