1. Decide which receipt fields you need
Start with the destination, not the OCR engine. A reimbursement workflow may only need merchant, date, total, currency, and payment method. A bookkeeping intake workflow may also need receipt number, subtotal, tax, and every line item.
ClearLedger's current structured fields are merchant, date, currency, receipt number, payment method, subtotal, tax or service fee, total, and line items with quantity, unit price, and amount.
A field should stay empty when the receipt does not support it. Requiring every column to be filled encourages guessed data.
2. Prepare the best available PDF
Use the original exported or scanned PDF whenever possible. Repeated screenshots, compression, skew, low contrast, and cropped edges make OCR and layout interpretation harder.
- Confirm that the receipt is readable at normal zoom.
- Remove unrelated pages if you are authorized to edit the file.
- Unlock the PDF before upload; ClearLedger rejects encrypted files.
- Keep the file under 20 MB and 30 pages.
3. Parse embedded text or scanned pages
A text-based PDF already contains selectable characters, so the parser can read them with their page coordinates. A scanned receipt is an image inside a PDF; it needs OCR before fields can be identified.
ClearLedger chooses per page: it reads an adequate text layer directly and uses OCR when the page text is too sparse. The result includes page text, structured fields, confidence, and source coordinates.
4. Review important fields against the source
Review is not merely proofreading spelling. You are checking whether the parser chose the right semantic value from several plausible values on the page.
| FIELD | COMMON AMBIGUITY | REVIEW QUESTION |
|---|---|---|
| merchant | Brand, location, legal entity | Which name identifies the seller for your workflow? |
| date | Order, transaction, print, or settlement date | Which date should the destination system use? |
| tax | Tax, tip, service fee, surcharge | Does the label support the classification? |
| total | Subtotal, amount due, paid, balance | Is this the final transaction amount? |
| items | Wrapped text, modifiers, quantities | Does each description match its amount? |
In ClearLedger, selecting a field highlights its source on the first-page preview when coordinates are available. Confidence helps prioritize attention, but it is not a guarantee.
5. Reconcile subtotal, tax, fees, and total
Use the receipt's own labels and arithmetic. A common check is subtotal plus tax equals total, but discounts, tips, deposits, and service fees may change the equation.
- Add the line-item amounts and compare them with the displayed subtotal.
- Identify discounts or credits separately instead of forcing the math.
- Add tax and clearly labeled fees.
- Compare the result with the final paid or due amount.
- Investigate differences larger than normal displayed rounding.
If the receipt remains ambiguous, flag it for manual follow-up. A visibly unresolved difference is safer than a fabricated correction.
6. Export and validate the CSV
After corrections are applied in the active page, export UTF-8 CSV before navigating away or refreshing. Open it in the destination spreadsheet or import tool and inspect several rows before processing a larger batch.
- Confirm column mapping and date format.
- Keep amounts numeric rather than adding currency symbols to value cells.
- Check multilingual merchant and item names for encoding.
- Retain the original receipt according to your own recordkeeping policy.
- Delete the parsed result from ClearLedger when you no longer need it.
For the exact current output fields, see the receipt-to-CSV workflow.
When manual extraction is still better
Manual entry may be the safer choice for a one-off receipt, severe handwriting, a damaged scan, or a document whose classification requires professional judgment. For repeated, readable PDFs, a parser plus review usually removes more repetitive typing while keeping a person responsible for the final data.