FY2026 EBITDA is $303K in Current against $452K in Prior, a change of ($149K), or (32.9%). Three of the six drivers moved. The script's walk adds up exactly to the total change, and every driver that didn't move shows $0.

**The walk, in the model's fixed order:**

| Step | Amount |
| --- | --- |
| Prior EBITDA | $452K |
| Active customers | ($133K) |
| Revenue per customer | $0 |
| Gross margin | $39K |
| Headcount | ($54K) |
| Cost per head | $0 |
| Other opex | $0 |
| Current EBITDA | $303K |

- **Active customers is the main driver.** It went from 46 to 44 in all 12 months, which takes ($133K) off EBITDA. It is also the whole revenue change of ($509K), or (4.3%).
- **Gross margin offset part of the drop.** It rose from 26.2% to 26.6% in all 12 months, adding $39K.
- **Headcount added cost.** It went from 23.00 to 23.75, changing in 9 of the 12 months, which takes off ($54K).

Revenue per customer, cost per head and other opex did not change.

The data shows how much each driver moved EBITDA, not why it moved. So the Review section of the note asks each driver's owner which documented assumption supports the change. For example, which customers are behind the drop in active customers, and whether there's an approved hiring plan behind the headcount change.

Files are in `output/`:
- drivers-note.md
- drivers.json