I wrote the memo to `FY2026_scenario_memo.md`. Base operating profit is **303,241**, a 2.71% margin. Upside reaches **500,572**, which is 197,332 above Base. Downside ends at a loss of **(6,147)**, which is (309,388) below Base.

- **Upside:** two Enterprise wins add 135,159 and the 2% January price rise adds 62,173. Costs don't change.
- **Downside:** three lost credit-watch accounts cost (202,738), the vendor cost that isn't passed on costs (78,270), and fuel surcharges cost (28,380). Lost customers make up 66% of the Downside.
- **Main point:** each customer is worth about 67,580 of gross profit. Customer count moves the result far more than any cost line, and headcount stays the same in all three scenarios.

**Open questions in the memo:**
- **Timing:** every driver is a full-year figure. The CSV doesn't say whether the January and March changes were pro-rated for the months they apply. If they weren't, the Upside and the margin part of the Downside are overstated.
- **Actuals:** it's October, so year-to-date actuals would show which scenario is happening.
- **Probabilities:** the file gives none for any scenario.

The figures come from `scenario_calc.py`, which I added. You can run it again with `python3 scenario_calc.py`. I checked by hand that each bridge adds up to its scenario's total. For style, I followed the house writing rules in the repo's close-reconciliation skill folder: lead with the result, and show negatives in parentheses. I didn't use the skill itself because it doesn't cover forecasts.