Cutoff: May 7 (latest data) − 8 days =
Apr 29 — days ≤ Apr 29 use actual conversions; days > Apr 29 use a fixed reference rate.
Fixed conv. rate window: Apr 23–Apr 29 → 45 conv / 197 starts =
22.8% (same rate applied to every projected day).
Week of Apr 26 | actual starts: 193 | projected upgrades: 42.7 | upgrade rate: 22.8%
4 days settled (≤ Apr 29), 3 projected using 22.8% rate.
| Day | Starts | Upgrades method | Conv. rate | Upgrades |
|---|
| Sun Apr 26 | 22 | actual conv. | — | 9.0 |
| Mon Apr 27 | 52 | actual conv. | — | 8.0 |
| Tue Apr 28 | 28 | actual conv. | — | 5.0 |
| Wed Apr 29 | 31 | actual conv. | — | 7.0 |
| Thu Apr 30 | 22 | rate × starts | 22.8% | 5.0 |
| Fri May 1 | 22 | rate × starts | 22.8% | 5.0 |
| Sat May 2 | 16 | rate × starts | 22.8% | 3.6 |
Fixed rate window (Apr 23–Apr 29) is the last 7-day period where all conversions are fully matured. Using a sliding window for post-cutoff days artificially depresses the rate as it draws in recent starts with no conversions yet.