Prepare the weekly revenue update

A weekly number that changes and is not explained gets ignored by the third week. The explanation is the deliverable; the numbers are how you find it.

The numbers, from one source each

Always the same source for the same number, or the series is not comparable week to week.

• Recurring revenue at the end of the week, and the change from last week.
• New: accounts started, and what they are worth.
• Expansion and contraction: upgrades and downgrades, by value.
• Churned: accounts ended and value lost, with each one named if there are few enough.
• Failed payments outstanding, which is revenue at risk rather than revenue lost.
• One usage number that leads revenue — active accounts, or whatever the product's real measure of use is — from the analytics or warehouse side, not from billing.

Then explain the movements

For every change larger than the week's noise, name the cause: which account, which deal, which price change. A figure with no cause attached should be checked before it is published — most unexplained jumps are a data problem, a duplicate, or a test account that got into the query.

Compare against the previous four weeks, not against the previous week alone. One week is indistinguishable from noise.

Publish

One Slack message: the headline number and its change, then five or six lines each with a number and its reason, then a link to the full breakdown. Charts go in the linked view, not in the message.

Stop and ask

• Before publishing a number that contradicts last week's without a known cause. Find out first.
• Before naming a customer's churn in a wide channel, if that fact is not already known there.