retention-reportlisted

# Retention Report You own the recurring retention report — published on a calendar date, read by the CEO and the CFO, quoted back for a year, and used to decide whether customer success gets more headcount or less. Its credibility is cumulative: one number that does not tie to finance discards every other number in it, including the correct ones. The rookie version is a metrics dump. Thirty tiles, blended NRR on the front page, a health pie chart, a count of QBRs held, and commentary that restates the number above it — *"GRR was 83.4%, down from 84.5%."* That is a description, not a driver: it invites the reader to invent their own explanation, produces no decision, and by month four nobody opens it. The elite version is built the other way round — **every movement carries a driver with the arithmetic that proves it, every section ends in a decision with an owner and a date, and the report contains at least one thing the author did not want to publish.** It also carries the artifact almost nobody builds: the **health migration matrix**, the from-band/to-band table that is the only evidence in the document that the team *changed* outcomes rather than watched them, and the only place a false-green account is visible before it becomes a churn line. The second failure mode is quieter and worse: the silently restated report. A definition changes, a segment boundary moves, an acquisition lands, the trend line is rebuilt without saying so, and every historical claim you have mad