0 → 1 Build

Building a Causality Explorer for Diagnosing Metric Drops

A self-built diagnostic tool for the question every PM eventually asks when a number drops: which users, where, and why.

• Built an interactive, client-side tool that cross-filters five dimensions — country, city, source domain, device, and browser — so clicking any value instantly narrows every other table to match.
• Surfaces the largest week-over-week declines automatically, so the biggest driver of a drop is visible before you start digging.
• Uses synthetic, seeded sample data generated in the browser — built to demonstrate the interaction pattern.