Read this before any specific number. It prevents most of the errors the rest of the measurement topics would otherwise walk into: the wrong denominator, the mix shift, the average that hides two populations, and the metric nobody can attach a decision to.
| # | Lesson | The question it answers | |
|---|---|---|---|
| 01 | What a Metric Is For ✅ | Which decision does this number change, and if none, why is it on a dashboard? | |
| 02 | The Denominator Problem ✅ | Why is conversion rate three different numbers, and which did you mean? | |
| 03 | Rates, Ratios and Averages ✅ | Why is the average nearly always the wrong summary? | |
| 04 | Windows, Cohorts and Snapshots ✅ | Why do two correct answers disagree about the same week? | |
| 05 | Simpson’s Paradox and Mix Shift ✅ | How can every segment improve while the total gets worse? | |
| 06 | Leading, Lagging and Vanity ✅ | How do you tell a predictive number from a flattering one? | |
| 07 | Benchmarks, Targets and Moving Goalposts | What happens to a metric once a target is attached to it? | |
| 08 | Survivorship and Selection | Which population is actually in your numerator? | |
| 09 | The North Star and Its Critics | What is a single top metric good for, and where does it mislead? | |
| 10 | Guardrails and Counter Metrics | What stops a team optimising one number into a worse product? | |
| 11 | Instrumentation: Getting the Number at All | What has to be true before a metric exists? | |
| 12 | Worked: Two Dashboards That Disagree | One real discrepancy, diagnosed to its cause |