Which receipt files can I upload on the web?
The current web uploader accepts one PNG, JPG, JPEG, GIF, WEBP, or PDF file at a time. File handling and extraction can still vary by document.
OCR receipt scanning workflow
Scan receipts with OCR, then review the details that matter
Expense Atlas uses OCR and assisted extraction to structure fields from supported receipt images and PDFs. You can inspect and correct the merchant, date, totals, payment details, categories, and line items before relying on the result.
From OCR receipt upload to reviewed expense record
Extraction reduces manual entry, but it does not remove the review step. Document quality, layout, available fields, and processing behavior all affect the result.
- 1
Upload a supported file
The current web upload accepts one PNG, JPG, JPEG, GIF, WEBP, or PDF file at a time. Clear, complete documents usually provide better extraction inputs.
- 2
Extract available fields
Processing can identify fields such as merchant, transaction date, subtotal, tax, tip, total, payment details, and individual line items when they are present and readable.
- 3
Review the receipt record
Open the resulting receipt and compare it with the source. Check totals, line items, dates, merchant normalization, and any fields that were not available.
- 4
Correct or complete details
Edit incorrect or missing values before using the record for categorization, reporting, or a budget input.
- 5
Review matching and categories
Expense Atlas can attempt categorization and supported reconciliation with imported activity. A suggested category or match can still be wrong, absent, or duplicated.
Synthetic demo data
Correcting a restaurant receipt before using it
A low-contrast receipt can produce useful fields and still need a human correction.
Extracted merchant
Matches the receipt header.
Harbor Cafe
Extracted subtotal
Matches the printed subtotal.
$42.00
Extracted tip
Handwritten tip was not captured.
$0.00
Corrected tip and total
User verifies the source and updates both fields.
$8.00 and $54.20
The corrected receipt can now support expense review. Any category or bank-transaction match still needs confirmation.
CSS illustration with synthetic demo data. This is not a product interface or screenshot.
Receipt extraction varies by document
- Blurry, cropped, dark, faded, handwritten, damaged, unusually long, or complex receipts can produce missing or incorrect fields.
- A supported file type does not guarantee that every layout, language, currency, or line item will be interpreted correctly.
- Password-protected, illegible, or non-receipt documents may fail or require a different input.
- Extraction, merchant normalization, categorization, and reconciliation are separate steps. Success in one does not guarantee success in the others.
- Monthly receipt capacity depends on the current plan. Review pricing for current limits before relying on a high-volume workflow.
Frequently asked questions
Clear answers about how this feature works and what to review before relying on it.
Which fields can OCR receipt scanning extract?
Available results can include merchant, date, subtotal, tax, tip, total, payment details, and line items. Fields that are missing, unreadable, or unusual may be absent or incorrect.
Is receipt extraction guaranteed to be accurate?
No. Results depend on image quality, document format, visible fields, and processing behavior. Review and correction are required for reliable expense records.
Will every receipt match a bank transaction?
No. Matching depends on available dates, amounts, merchant details, and imported activity. Review suggested matches and duplicates before relying on them.