How to extract data from handwritten receipts
Dynamite Docs, 2026-08-30
Capture a readable image
Place the receipt on a flat, contrasting surface and photograph it from directly above. Include every edge. Use even light and check the image before discarding the paper.
A higher-resolution image cannot recover writing hidden by a fold, finger, shadow, or torn corner. Take a second photo if the merchant, date, or total is hard to read with your own eyes.
Extract only fields supported by the receipt
A practical expense schema may include merchant or payee, date, description, subtotal, tax, tip, total, currency, payment method, and project or employee. Many handwritten receipts omit some of these fields.
Leave an absent field blank. Do not infer a tax amount by subtracting plausible values or turn an unreadable merchant name into a confident guess.
Treat handwriting confidence field by field
One receipt can contain a clear printed date, a faint handwritten description, and an overwritten total. Give each field its own confidence and review state. A single document-level score hides the exact cells that need attention.
Amounts deserve priority because a missing decimal point changes the expense. Dates, merchant names, and approval signatures may also affect reimbursement rules.
Separate subtotal, tax, tip, and total
Restaurant and taxi receipts often have a printed subtotal with handwritten tip and total. Capture each as a separate value. Compare subtotal plus printed tax plus tip with the handwritten total when all parts are readable.
If the writer changed the tip or total, keep the final visible value and flag the overwrite for review.
Keep the image linked to the expense row
Store a document ID or source link with the extracted row. A reviewer should be able to open the receipt without searching through a camera roll or email thread. Apply your retention policy after reimbursement and audit needs are met.
Review handwritten receipts in Dynamite Docs
Upload receipt photos or a scanned PDF. Dynamite Docs uses visual extraction for handwriting, shows the source beside the result, and flags uncertain fields. Correct the few cells that need human judgment, then export the approved expenses to Excel, CSV, or Google Sheets.
Related workflow: Receipt data extraction and OCR for real-world receipts.
Try it yourself. Upload a PDF, scan, or image and let Dynamite Docs infer the schema.