Essay
Metrics product teams actually trust
Trust in a metric rarely comes from a prettier chart. It comes from a definition that survives the question “what would make this number wrong?” Product teams that skip that question end up defending vanity counters because nobody wrote down the exclusions.
Start with the decision, not the warehouse
Before you rename columns, list the decisions the metric is supposed to change. If the answer is “keep an eye on things,” you are building ambient anxiety, not a reporting system. In our Product Reporting Systems cohorts we ask pairs to name one decision per metric — hire a contractor, kill a experiment, change onboarding copy — and discard anything that cannot claim one.
Publish the weaknesses
Trusted metrics admit their blind spots. A retention curve that excludes a known tracking gap should say so in the footnote of every pack. Stakeholders forgive provisional labels; they punish silent surprises.
Retire with ceremony
Deleting a popular vanity metric without a short note invites revival next quarter. Write a one-paragraph retirement memo: what it measured, why it misled, what replaces it. Archive the chart; keep the memo in the metric contract.