Magazine apps taught a generation of analysts to think in calendar days: D1, D7, D30. Those windows still fit a news brief. They fit a monthly billed tool the way a bus timetable fits a ferry. The object that matters is the billing period, the pause, and whether a job was completed inside the period you charged for.
Retention Cartography starts by drawing the window on paper before anyone opens a vendor. For a monthly plan we want: first paid period, first completed job inside that period, and whether the second invoice was attempted. D7 of “opened the app” can look healthy while the second invoice fails because the card expired and nobody built the dunning event.
Pauses are the other country. A pause is not churn until the billing object says the seat is gone. Teams that file pauses as churn make resurrection look miraculous when the user simply came back from August. The Atlas keeps pause on plate three, resurrection on plate four, and refuses the average.
Annual plans need a longer patience than a growth slide likes. We still look at an early job — did they finish onboarding in the first fortnight — but we do not call a quiet month in a yearly seat “death.” Quiet can be the product working. A design tool opened twice a month may be earning its invoice.
If you take one thing from the room, take this: print the window on the chart. “Cohort of March paid starts, observed through the second invoice.” That caption does more work than a gradient. Growth may still want D7 for a campaign. Let them have it, labelled as campaign residue, not as loyalty.