# Northline Workshops — October 2026 reconciliation

## Basis and headline results
All amounts are CAD, tax excluded, for October 1–31, 2026. Paid gross less eligible settled October refunds is net revenue; contribution after advertising is net revenue less attributable October ad spend. These definitions and the exclusion rules follow `inputs/policy.md`, **Northline Workshops accounting note — October 2026**.

| Metric | Workshop | Course | Combined |
|---|---:|---:|---:|
| Paid order count | 5 | 4 | 9 |
| Gross sales | 1,920.00 | 900.00 | 2,820.00 |
| Applied refunds (known, eligible amounts) | 240.00 | 60.00 | 300.00 |
| Net revenue | 1,680.00 | **840.00 provisional** | **2,520.00 provisional** |
| Advertising spend | 300.00 | 150.00 | 450.00 |
| Contribution after advertising | 1,380.00 | **690.00 provisional** | **2,070.00 provisional** |
| Net revenue per paid order | 336.00 | **210.00 provisional** | **280.00 provisional** |

Source rows: Workshop sales `inputs/sales.csv` S01, S03, S06, S12, S14; Course sales S02, S07, S10, S13. Applied refunds: `inputs/refunds.csv` R01 (Workshop), R02 (Course). Ads: `inputs/ads.csv` A01, A03, A06 (Workshop) and A02, A05 (Course).

**Do not treat the provisional values as final.** R06/RF05 in `inputs/refunds.csv` is settled in October against paid Course order N1011 (S13) but has no amount. Its blank is unknown, not zero. R05/RF04 has a known 120.00 amount but references nonexistent order N9999; it is not applied or assigned a product. Resolve that unmatched refund as well before closing the accounts. Neither issue has been guessed away.

If the missing R06 amount is x and no other records change, Course net revenue is 840.00 − x and contribution is 690.00 − x; combined net revenue is 2,520.00 − x and contribution is 2,070.00 − x. The corresponding per-order figures are (840.00 − x) / 4 and (2,520.00 − x) / 9. These formulas do not assume how R05 will ultimately be resolved. [Sources: `inputs/refunds.csv`, R05–R06; `inputs/sales.csv`, S13; `inputs/policy.md`, **Northline Workshops accounting note — October 2026**.]

## Calculation and counting rules
- Count unique included paid orders, not units, raw rows or orders net of refunds. The denominator remains nine paid orders, including refunded orders. Net revenue per paid order = net revenue / paid order count.
- Workshop: 1,920.00 − 240.00 = 1,680.00 net; less 300.00 ads = 1,380.00 contribution. Course calculable total: 900.00 − 60.00 = 840.00 net; less 150.00 ads = 690.00 contribution. Combined: 2,820.00 − 300.00 = 2,520.00 net; less 450.00 ads = 2,070.00 contribution. Source row groups are listed above.
- Choose the latest `updated_at` for conflicting order versions: S10 replaces S09 for N1008, so gross is 180.00, not 120.00 or 300.00. Duplicate business records differing only in source `row_id` are collapsed by retaining the first occurrence: S03 over S04, R02 over R03, A03 over A04. This is the operational interpretation of “exact duplicate rows,” since source row IDs are distinct. [Sources: `inputs/sales.csv`, S03–S04, S09–S10; `inputs/refunds.csv`, R02–R03; `inputs/ads.csv`, A03–A04; policy heading above.]
- Settlement date controls refund inclusion. The 240.00 November refund on October order N1012 is excluded from October; the pending refund on N1005 is not applied. [Source: `inputs/refunds.csv`, R04, R07; policy heading above.]

## Record reconciliation
All raw rows remain in the cleaned files for audit, including excluded rows. `included_october=yes` selects the calculation population; `no` means excluded from the reported October total, not deleted from the source. Every excluded row has one primary reason in `exceptions.csv`.

| Dataset | Source rows | Included rows | Excluded rows | Included source row IDs |
|---|---:|---:|---:|---|
| Sales | 15 | 9 | 6 | S01, S02, S03, S06, S07, S10, S12, S13, S14 |
| Refunds | 7 | 2 | 5 | R01, R02 |
| Ads | 7 | 5 | 2 | A01, A02, A03, A05, A06 |
| Total | 29 | 16 | 13 | Counts are records across different datasets, not orders |

Exclusion bridge (source filenames identify the underlying evidence):

| Source | Excluded row(s) | Reason |
|---|---|---|
| `inputs/sales.csv` | S04 | Duplicate of S03 |
| `inputs/sales.csv` | S05; S08 | Pending; cancelled |
| `inputs/sales.csv` | S09 | Superseded by S10, latest update |
| `inputs/sales.csv` | S11; S15 | September order; November order |
| `inputs/refunds.csv` | R03 | Duplicate of R02 |
| `inputs/refunds.csv` | R04 | Pending, not settled |
| `inputs/refunds.csv` | R05 | No matching paid order N9999 |
| `inputs/refunds.csv` | R06 | Missing amount; retained blank |
| `inputs/refunds.csv` | R07 | November settlement |
| `inputs/ads.csv` | A04 | Duplicate of A03 |
| `inputs/ads.csv` | A07 | November spend |

## File conventions and assumptions
The three cleaned CSVs retain all original source columns and row IDs, with normalized values and inclusion/reason columns appended. Valid date-only fields use YYYY-MM-DD; `updated_at` retains its full original ISO timestamp so version selection remains auditable. Money has two decimal places except the unresolved blank R06 amount. Products use Workshop/Course; ad channels are lowercase and trimmed. Refund product is derived only from a matched paid order; R05 remains blank, not assigned. No unmapped supplied product was encountered. [Sources: all rows of the three input CSVs; normalization rules in `inputs/policy.md`, heading above.]

Excluded rows remain normalized but do not enter totals. No source input has been edited. Known applied refunds were checked cumulatively per order against paid gross; no excess was found among the calculable included refunds. The unknown R06 amount cannot be cap-checked until supplied. [Sources: `inputs/refunds.csv`, R01–R07; matched paid records in `inputs/sales.csv`.]

## Recommendation and limitations
First obtain the settled amount for RF05/R06 and investigate RF04/R05 against the order ledger; do not finalize Course or combined net revenue until those issues are resolved. Keep RF06/R07 for the November settlement reconciliation. [Source: `inputs/refunds.csv`, R05–R07.]

Workshop currently contributes 1,380.00 after ads versus Course's provisional 690.00. Maintain product-level tracking and use only a cautious, measured advertising test rather than assume that more spend will reproduce these results: the supplied rows support attribution of October ad spend by product, not causal return or future performance. **This contribution is not business profit.** COGS, payment fees, salaries and overhead are absent, so their impact cannot be assessed. [Sources: calculation table and row groups above; `inputs/policy.md`, **Northline Workshops accounting note — October 2026**.]

## Checks performed and handoff
Local Python standard-library checks passed for expected product and combined metrics; included counts (9 sales, 2 refunds, 5 ads); uniqueness of included entity IDs; all 13 exclusion links; cumulative known applied refunds not exceeding paid gross; cleaned CSV write/read equality; date and money formats; preservation of the R06 blank; and unchanged SHA-256 hashes of all input files during processing. These are internal consistency checks against the supplied fixtures, not independent verification against a bank or external ledger.

Deliverables: `output/cleaned-sales.csv`, `output/cleaned-refunds.csv`, `output/cleaned-ads.csv`, `output/exceptions.csv`, and `output/summary.md`.

Remaining limitations: R06 is missing its amount; R05 lacks a matching order; expense categories needed for profit are not supplied. Estimated human effort: **10–20 minutes** to review and update the reconciliation once supporting refund/order records are available; obtaining those records may take additional, unknown time. This is an estimate, not measured work time.
