I've written the FY2026 scenario memo to `output/scenario-memo.md`. Base EBITDA is $303K, with a range from ($6K) in Downside to $501K in Upside. The script checked that each scenario's driver changes add up to its EBITDA to the cent, and they all do. Every figure in the memo is copied from its output, `output/scenarios.json`.

| Scenario | Revenue | EBITDA | vs Base |
| --- | --- | --- | --- |
| Base | $11.2M | $303K | n/a |
| Upside | $11.9M | $501K | $197K, 65.1% |
| Downside | $10.4M | ($6K) | ($309K), (102.0%) |

**What drives the difference.** Each driver's figure is how much it moves EBITDA from Base. The drivers are counted in a fixed order, so these figures aren't each driver's share of the total.
- **Upside:**
  - Active customers go from 44 to 46 if 2 Enterprise proposals close in January: $135K.
  - Revenue per customer rises if the January price increase sticks without losing customers: $62K.
- **Downside:**
  - Active customers fall from 44 to 41 if the 3 accounts on the credit watch list are lost: ($203K).
  - Gross margin falls from 26.6% to 25.8% if the March vendor cost increase isn't passed on to customers: ($78K).
  - Other opex rises from $946K to $974K on higher carrier fuel surcharges: ($28K).
- Headcount and cost per head are the same in all three cases.

**Recommendation in the memo:** plan to Base. The signals to watch are:
- whether the Enterprise deals sign in January;
- whether customers accept the price increase;
- whether any credit watch list accounts leave;
- whether the vendor cost increase can be passed on.

The memo suggests preparing a plan to pass on the vendor cost increase, plans to keep the credit watch list accounts, and a list of opex cuts to fall back on.