I've produced the five October files, but the results are provisional: one settled refund (RF05 on order N1011, a Course order) has no amount in `refunds.csv`. Following the policy, I left it out rather than treating it as zero.

**October 2026 results (CAD, tax excluded)**

| | Workshop | Course | Combined |
|---|---:|---:|---:|
| Paid orders | 5 | 4 | 9 |
| Gross sales | 1,920.00 | 900.00 | 2,820.00 |
| Refunds applied | 240.00 | 60.00 | 300.00 |
| Net revenue | 1,680.00 | 840.00 | 2,520.00 |
| Ad spend | 300.00 | 150.00 | 450.00 |
| Contribution after advertising | 1,380.00 | 690.00 | 2,070.00 |
| Net revenue per paid order | 336.00 | 210.00 | 280.00 |

- **Effect of the missing refund:** a refund can't exceed the order's gross of 360.00, so this is a range, not an estimate. Course net revenue could be 480.00–840.00. Combined net revenue could be 2,160.00–2,520.00 and combined contribution 1,710.00–2,070.00. Workshop figures aren't affected.
- **Reconciliation:**
  - **Sales:** 15 rows; 9 included (2,820.00) and 6 excluded (1,320.00), totalling 4,140.00.
  - **Refunds:** 7 rows; 2 included (300.00) and 5 excluded (660.00 plus the one unknown amount).
  - **Ads:** 7 rows; 5 included (450.00) and 2 excluded (180.00), totalling 630.00.
- **Exclusions:** duplicate rows, pending or cancelled orders and refunds, and records dated outside October. One superseded version of order N1008 is also excluded; the later version (180.00) is counted. RF04 (120.00 settled) is excluded because it points to order N9999, which doesn't exist as a paid order. Each excluded row and its reason is listed in `exceptions.csv`.
- **Recommendation:**
  - Get the RF05 amount and find out what RF04 relates to before closing the month.
  - Both products returned 5.60 of net revenue per ad dollar. Workshop earns more per order (336 vs 210) and brings in two-thirds of the contribution. Nine orders in one month is too little to justify moving ad budget between products.
- **Not profit:** the 2,070.00 contribution doesn't deduct delivery costs, payment fees, salaries or overhead, because none were supplied. Actual profit will be lower by an unknown amount.

**Assumptions:**
- The cleaned files keep every source row, with "included in totals" and "exclusion reason" columns. Filter on the first to get only the counted records.
- Orders count in the month of their order date, refunds in the month they settled, and ads in the month of their date.

**Checks performed:** I ran `build_task1.py` and it checks all 21 metrics against my own hand calculation; all matched. I also hand-checked that included plus excluded rows and amounts equal the raw totals in all three files, and read through `exceptions.csv`, `cleaned-refunds.csv` and `cleaned-ads.csv`. I didn't open `cleaned-sales.csv` to read it, though the script's checks cover its totals.

**Remaining limitations:** RF05's amount and RF04's real order are still unresolved. No cost data was supplied, so profit can't be calculated.

**Estimated time for a person to finish:** about 15–30 minutes to review, plus however long it takes to get the RF05 and RF04 details from the payment records.

Files are in `output/`:
- `cleaned-sales.csv`
- `cleaned-refunds.csv`
- `cleaned-ads.csv`
- `exceptions.csv`
- `summary.md`
- `build_task1.py`