Journal
Cohort curves that travel between teams
Cohort curves die in handoffs. The original author remembers which filters were temporary; the next owner treats them as law. Good business analytics for revenue forecasting treats the curve as a document with a legend, not a private artefact.
Label the intake rule
State whether a customer enters the cohort on contract date, first invoice, or first usage. GB teams often disagree here because finance and sales ops run different clocks. Write the rule above the chart, not in a buried comment.
Separate shape from level
Shape describes how revenue or retention unfolds over months. Level is the starting size. When you update a curve, note which of the two moved. Mixing them creates phantom improvement.
Version lightly
Keep the previous curve visible for one cycle with a grey overlay. Stakeholders learn faster when they can see the edit, not just the new truth. In Forecast Studio we practice this as a standing habit rather than an annual cleanup.
If your data warehouse cannot support clean cohort keys yet, publish a smaller trusted set instead of a comprehensive fiction. Travel-ready curves beat ornate ones that only one person can defend.