Product Analytics
Metric design, funnels, retention cohorts, event schemas.
The North Star metric: the one number a whole team rallies around
A North Star metric is the single number a company trusts most to mean 'the product is genuinely getting better.' Most teams accidentally pick the wrong one.
Retention curves & cohort tables: the chart that tells you if a product really works
One retention number tells you almost nothing. Grouping users by when they joined, and tracking each group over time, is the closest thing analytics has to a truth serum.
Funnel decomposition: finding exactly which step is broken
One overall conversion rate tells you something's wrong. Breaking it into step-by-step numbers tells you what, and that's the entire skill.
DAU, MAU, and stickiness: simple math with three hidden traps
Dividing DAU by MAU is one line of SQL. Getting a number that actually MEANS something takes knowing three things that go wrong first.
Event tracking schemas: the decision that decides what questions you can ever answer
A messy event schema doesn't cause bugs today. It causes questions six months from now that simply can't be answered, no matter how good your SQL is.