# Review change log

## Changes
1. **Preserved the original submission** in `output/original-submission/`: all five original deliverables, `reconcile.py`, and `sha256.json`. The archive was created before revisions, and its file hashes were verified afterward.
2. **Corrected an evidence overstatement** in `summary.md`: changed “nonexistent order N9999” to an order with no match in the supplied sales file. The evidence establishes absence from `inputs/sales.csv` S01–S15, not nonexistence in an external ledger. R05 remains excluded under `inputs/policy.md`, **Northline Workshops accounting note — October 2026**.
3. **Added monetary reconciliation controls** to `summary.md`: raw-row sales 4,140.00 = 2,820.00 included + 1,320.00 excluded; known refunds 960.00 = 300.00 included + 660.00 excluded, with one additional unknown amount retained; ads 630.00 = 450.00 included + 180.00 excluded. Sources: `inputs/sales.csv` S01–S15, `inputs/refunds.csv` R01–R07, `inputs/ads.csv` A01–A07.
4. **Added a review audit record** in `review-checks.txt` and clarified review scope in `summary.md`.

## Unchanged results and files
All 21 headline values were independently recomputed from raw records and agree with the first submission. The four CSV deliverables remain byte-identical to the archived originals: `cleaned-sales.csv`, `cleaned-refunds.csv`, `cleaned-ads.csv`, `exceptions.csv`.

Combined results remain 9 paid orders, CAD 2,820.00 gross, CAD 300.00 applied refunds, CAD 2,520.00 provisional net revenue, CAD 450.00 ad spend, CAD 2,070.00 provisional contribution after advertising, and CAD 280.00 provisional net revenue per paid order. Product detail and record citations remain in `summary.md`.

## Checks actually performed
- Independently selected eligible records from raw dates, IDs, statuses and latest sales timestamps; confirmed the sole conflicting order is N1008 (S09/S10) and that S10 is latest.
- Verified all 29 raw row IDs, all original fields against normalized values, 16 inclusions, 13 one-to-one exception links and unique included entity IDs.
- Checked CSV structure, date-only formats, preserved update timestamps, two-decimal known amounts, normalized products/channels, and preservation of the unknown R06 amount.
- Verified matched refund products, individual known matched refund amounts against gross, and cumulative applied October refunds against gross.
- Recomputed product/combined metrics and raw-to-included/excluded monetary controls.
- Verified unchanged input hashes during review and the original archive manifest; confirmed unchanged CSV deliverables by hash.

These checks used Python standard library and the supplied fixtures only; no bank, processor, external ledger, spreadsheet-application or browser testing was performed. The scripts establish correctness for these fixtures, not a general-purpose ingestion system for unseen data.

## Handoff
Current deliverables: `output/cleaned-sales.csv`, `output/cleaned-refunds.csv`, `output/cleaned-ads.csv`, `output/exceptions.csv`, `output/summary.md`. Review additions: `output/change-log.md`, `output/review-checks.txt`. Original submission: `output/original-submission/`.

Remaining limitations: R06 lacks its settled amount, R05 has no matched supplied order, and business-profit costs are not supplied. See `inputs/refunds.csv` R05–R06 and `inputs/policy.md`, **Northline Workshops accounting note — October 2026**. Estimated human follow-up is **10–20 minutes once supporting refund/order evidence is available**, plus unknown time to obtain it. This is an estimate, not a measured result.
