The 71% figure is a single-cohort read from one quarter, and it has been quoted since as if it were the account-wide rate.
Demo — not real data. Seeded, internally consistent, matched to the real corpus shape; nothing here is a live query.
The number originates in a single retention read taken on 2024-11-21the only source that computes it, covering the cohort that renewed inside the pilot window only — 71.4% across n=138. put it in the playbook deck the same week, and every later mention traces back to that deck rather than to a fresh calculation.
Two things have changed since. The renewal cycle was restructured in Apr 2025, which moved roughly a third of accounts out of the cohort the figure was computed over; and the account itself has grown past the population the pilot sampled. The Staleness lens flags this fact as past its domain half-life: it is reported as a historical reading, not asserted as the current rate.
There is no superseding measurement in the Well. The most recent atom that touches retention is an action row assigned to to re-run the cohort, still open. Until that lands, any external use of 71% needs the cohort and the date stated alongside it.
Pilot cohort retention lands at 71.4% (n=138) against a 64% target. Caveat for the deck: this is the pilot window only — the account-wide figure is not computed here.well://source/S-028#chunk-4
- Is the account-wide retention rate ever going to be computed, or is the pilot cohort the standing definition?
- Does the Apr 2025 renewal restructure invalidate the comparison, or just narrow it?