Topic
Metric best practices
Tool-agnostic rules for metrics you can trust. Each guide states the rule, shows how to check your own model, and points to where it went wrong in a real project.
Rules that hold on any stack
Why do totals go up after a join?Preventing join fanout by declaring every relationshipBest practicesRevenueWhy do two reports disagree on the same metric?Filters that quietly change what a metric meansBest practicesAcquisitionRetentionCan you divide clicks by impressions from different tables?Calculating CTR from separate impression and click tablesBest practicesAcquisitionWhat is this count actually counting?Checking the primary key before you trust a countBest practicesRevenueRetentionCan you trust an id extracted from a string?Why parsing ids out of strings at query time is riskyBest practicesAcquisitionWhy does a model look fine until someone runs it?Catching missing tables and stale references before a dashboard breaksBest practices