ACME 1,000 Receipts: Extract Transactions and Explain Adjustments

The challenge

You've got 1,000 receipts and no spreadsheet. Build a sales ledger, then investigate the receipts whose totals don't quite match their items.

Start here

Open the SharePoint folder linked above. receipts.pdf is the only input. It has 245 pages containing 1,000 receipts, with multiple receipts on most pages. The text is selectable, so a text extraction pass is an appropriate starting point.

Try it

1. Extract one transaction record and its item rows for each receipt, retaining page source details.

2. Calculate transaction count, recorded sales, item units, and the loyalty-marker rate.

3. Reconcile item subtotals and printed totals without replacing the source amounts.

4. Explain which identifiers and timing details cannot be recovered from these receipts.

What to build

Build transactions.csv and transaction_items.csv, or equivalent nested JSON, plus a reconciliation report with worked examples. Create local extraction IDs instead of inventing customer or transaction IDs from another dataset.