In this exercise, the interviewer plays a customer whose BI dashboard on Databricks SQL has become slow and expensive after data growth. You are given simplified query logs, table sizes, cache hit rates, concurrency levels, and warehouse configuration in JSON or CSV form. Write a program that analyzes the inputs and outputs the top likely bottlenecks, a prioritized remediation plan, and an estimated before/after impact on latency and cost using simple heuristics. Your code should surface issues such as skew, missing partition pruning, underprovisioned concurrency, or over-sized clusters, and rank recommendations by expected ROI. During the role-play, explain your findings as if speaking first to the customer’s data engineer and then reframe the same conclusions for a non-technical executive sponsor.