Analytics playbook

A field sequence we use with product teams: decide what matters, instrument it cleanly, validate the pipes, diagnose behavior, then make the practice stick.

Name the decisions

List the monthly product decisions that currently rely on anecdote. Each decision becomes a candidate for a metric or funnel — not a dashboard wish list.

Write the tracking plan

Define objects, actions, and properties before code ships. Owners approve the plan; engineers implement against acceptance checks, not vague tickets.

Validate the pipes

Confirm volume, null rates, and identity stitching in staging and production. Silent failures here invalidate every chart that follows.

Diagnose the paths

Rebuild funnels and retention with agreed definitions. Separate absolute counts from percentages so rising curves cannot hide shrinking bases.

Institutionalize the cadence

Daily pipeline health, weekly input metrics, monthly outcomes. Keep each layer short so analytics stays attached to shipping, not slide theater.

Put the sequence to work

Most teams begin with an Instrumentation Audit, then move into funnel diagnostics or a metrics framework once definitions hold.