Quick summary
Use multilingual OCR to digitise purchase orders and delivery notes, then compare product codes, quantities and receiving exceptions. This guide gives you a clear, practical explanation before you use the related online tool.
Why these documents are difficult
Purchase orders and delivery notes often use supplier-specific layouts, abbreviated descriptions, handwritten receiving notes and mixed product identifiers.
Prioritise stable identifiers
After extraction, match barcode, SKU, supplier item code or internal product code before comparing descriptions. Product names can differ while identifiers remain stable.
Normalise quantities and units
Convert obvious unit variations such as pcs, pieces and each into a controlled unit dictionary. Do not assume a carton equals a fixed quantity unless pack size is known.
Detect receiving exceptions
Compare ordered and delivered quantity to find shortages, over-deliveries, missing items, unexpected items and possible duplicates.
Handle uncertain OCR
A low-confidence digit can turn 10 into 70 or alter a product code. Route uncertain lines for manual verification instead of forcing an automatic match.
Create the receiving decision
Record accepted quantity, rejected quantity, reason and final decision: accept, partial accept, hold or reject. Preserve the original source for audit.
Continue with a free tool
Related FormatForge tools
Multilingual Document OCR
Extract editable text from scanned invoices, purchase orders and delivery notes using browser-based OCR for English, Greek, Hindi, Arabic, German, French, Spanish, Italian and Dutch.
Open tool →PO vs Invoice Matcher
Compare purchase-order and invoice files to detect missing lines, extra charges, quantity differences and price mismatches before payment.
Open tool →Three-Way Match Analyzer
Reconcile purchase-order, goods-receipt and supplier-invoice quantities and prices using configurable tolerances.
Open tool →Frequently asked questions
Can OCR compare a PO and delivery note directly?
OCR extracts text first; a matching engine then normalises and compares line items.
Does row order need to match?
No. Reliable comparison should use identifiers and normalised values, not row position.
Can handwritten notes be recognised?
Printed text is usually more reliable; handwriting quality varies significantly.
What should be manually checked?
Product codes, quantities, units, batch details and receiving decisions.
Keep learning
Related guides
OCR
How Browser-Based OCR Works: A Practical Guide
Understand how client-side OCR extracts text from PDFs and images, what happens in your browser, and where human verification is still required.
OCR
Is Online OCR Safe for Confidential Documents?
Learn how to evaluate OCR privacy, distinguish client-side processing from cloud uploads, and apply a practical security checklist.
OCR
OCR for Invoices: From Scanned PDF to Verified Data
A step-by-step invoice OCR workflow for extracting supplier details, invoice numbers, line items, taxes and totals while controlling errors.
OCR
Multilingual OCR Language Selection Guide
Choose the right OCR language model for English, Greek, Hindi, Arabic, German, French, Spanish, Italian and Dutch documents.