This month
Is the model right so far?
Monthly voluntary churn, all clubs
Scorecard
Treated against held-out is the only read that says whether the program is working. It fills in from the November 1 roster for members enrolled in October.
Where we stand
The two baseline numbers
What a normal month looks like once it runs
If the save rate is higher or lower
Pre-cancel outreach
Win-back thread
Value = members kept × 3 extra months × $33 dues. Revenue kept, not margin. Every figure on this tab except the churn history is a target until the October scorecard reads.
Data on disk
September 1 freeze, read against the mirror
Recall by list size
Tiers and deciles
Weekly scoring runs
merged_v1 run on each Monday with the check-ins to that date. Roster fields from the September 1 export (shifted so month keys line up); payments frozen at September 7. Contactable = visited in the last 15 days. Paused = 16 to 29 days since the last visit, goes to the dormant flow.
September rehearsal (simulated enrolment on real scores, visits and filings)
Monthly list against weekly scoring
Same model, same month. Weekly at 20% finds more filers by enrolling twice as many people; at 10% (about the same names as the monthly list) it finds fewer within September, partly because members enrolled late have had days, not weeks, to file. The October roster settles it.
New joiners and the Habit Builder gap
Enrolment ledger
By segment
Pipeline
Clubs, latest scores
Contactable and on-list from the September 22 weekly score (top 20% of contactable, fleet cut). RED from the September 1 freeze (within-club tiers). Filed = Pending Cancel or Cancelled in the mirror since September 1, among members active on September 1. Click a column to sort.