Weekly: what you can act on within the week. Monthly: what takes a month to move. About ten metrics in all, no more. The test we give every team fits in one picture: open a tab on Monday morning and see five numbers, with their change.
The benchmark
| Cadence | Metrics |
|---|---|
| Week | Human sessions, revenue per visit, conversion rate, average order value, orders, stockouts on flagship products, share of unattributed traffic |
| Month | Contribution margin, return rate, share of known customers, acquisition cost, repurchase rate, GA4 coverage rate |
| Quarter | Customer value by cohort, profitability by channel |
Pick the cadence based on how fast the number can move and on the decision it informs. At one clothing brand, the twelve-month repurchase rate fell to 20%: eight in ten customers bought only once. You don't read that number on Monday. It doesn't move from one week to the next, and it only matters if someone is there to open a CRM project.
The rule
Every metric has an owner and a decision attached to it. If nobody knows what to do when it moves, remove it from the report.
The steps
- The current list. For each metric, write down the decision it informs.
- The cut. Remove any metric that informs none.
- The definition. One written line for each, with its source: Shopify for orders, GA4 for channels.
- The point of comparison. Always two: last year and the target.
- The peak season. During the months that make the year, freeze the list at five numbers and ban any product decision other than removing what's broken. At one sports brand that makes 70% of its revenue between mid-November and the end of February, that's the rule written into the plan.
- The line for leadership:
"Five numbers on Monday, each with a name next to it."
The trap
Tracking weekly numbers that can't move in a week. A repurchase rate discussed every Monday produces invented explanations.
Conversely, finding out about a stockout on a flagship product in the monthly report means three weeks of lost sales.
