Fill baseline minutes, volume, rework, review rate, model cost, and platform cost for one loop. Payback is a date, not a slogan.
One loop only. Copy the row for a second loop. Do not average three processes into one heroic payback.
Use last-30-day counts you can show, not hoped hours. If a cell is a guess, write GUESSED next to it.
flowchart LR
B["Baseline minutes"] --> N["Net minutes after"]
C["Model plus platform"] --> P["Payback months"]
N --> P
R["Rework plus review"] --> N
| Cell | What to write | Do not write |
|---|---|---|
| Volume | Count of jobs in the window | A yearly hope divided by 12 |
| Human minutes each | Timed, or a sampled average | “It feels like an hour” |
| Rework rate | % of jobs that come back | Zero because you are optimistic |
| Minutes to fix a rework | Clocked | Ignored |
Net minutes saved × your loaded minute cost, minus model and platform cost. If that is negative, you bought a demo. Write the month you recoup build time, or write NEVER.