How to add document OCR to SAP Business ByDesign
Short answer
Run documents through an OCR and extraction service, validate against ByDesign business partners, purchase orders and goods receipts using web services, and create supplier invoice or receipt drafts. Anything below a confidence threshold goes to a review screen before it touches the ledger.
Accounts payable in a ByDesign company is often one or two people keying supplier invoices from PDFs, while receiving staff photograph packing slips and quality teams hold certificates in folders. Each paper or PDF holds structured information that ByDesign needs: supplier, invoice number, PO reference, tax, line amounts, batch and heat numbers.
OCR on its own only turns pixels into text. The valuable part is turning that text into a correct draft: matching the supplier, finding the PO, comparing quantities and prices to the goods receipt, and checking tax. That is where most failures happen, and where design effort pays off.
Since ByDesign is SaaS and exposes only supported services, the OCR pipeline lives outside it and posts drafts. That keeps upgrades painless and avoids modifications to the core system.
What teams use it for
Supplier invoices against purchase orders
Extract header and lines, find the PO, compare with goods receipt, and create a supplier invoice draft when within tolerance.
Packing slips to goods receipt
Photograph or scan a delivery note at receiving, read PO reference and quantities, and propose a goods receipt for the clerk to post.
Certificates of conformance and mill certs
Capture supplier, material, batch and test values; attach to the inbound delivery or lot and flag missing certificates before use.
Customer POs and drawings
Read customer order PDFs and attached drawings to create sales order drafts and link the drawing revision.
Expense receipts
Extract date, vendor, amount and tax from receipts for expense report lines, with policy checks on categories.
Customs and freight documents
Capture HS codes, values and container numbers for import shipments and link them to the related purchase orders.
SAP Business ByDesign integration surface
- Supplier invoice, purchase order, goods receipt (inbound delivery) and expense services reachable through Communication Scenarios and Arrangements, with names and fields depending on release.
- Custom web services in the Cloud Application Studio for attaching a source document or creating a document with only the fields your OCR provides.
- Business partner, material and tax code lookups via report feeds or query services, used in validation and fuzzy matching.
- Document attachment capability on business documents, so the original scan travels with the draft.
- ByDesign's own invoice and payment workflows, which still apply after your draft is created; your pipeline should not bypass approval rules.
How it works
- 1
Collect the documents
Take inputs from the AP mailbox, a scan folder, or a mobile upload. Store the original, hash it to detect duplicates, and record the source.
- 2
Run OCR and layout analysis
Use an engine suited to your documents: cloud services, or open-source plus a vision-language model for on-prem. Preserve page and coordinate data for the review UI.
- 3
Extract fields with confidence
Return structured fields and a confidence score per field. Totals, tax and line sums should be recomputed and cross-checked, not trusted.
- 4
Validate against ByDesign
Resolve the supplier by VAT or tax ID and bank details, find the PO by reference, and compare quantities and prices with goods receipts. Check for duplicate invoice numbers per supplier.
- 5
Route by confidence
High confidence plus a three-way match creates a draft automatically; anything else goes to a review screen showing the image beside the fields.
- 6
Post the draft and attach the file
Call the supplier invoice or goods receipt service with the extracted data, attach the original, and keep the ByDesign document ID against your record.
- 7
Feed corrections back
Save supplier layouts, aliases and unit mappings to lift accuracy on repeat senders.
Example flows
Supplier invoice with a price variance
Trigger: A PDF invoice arrives in the AP inbox.
- OCR extracts header and four lines with high confidence.
- The matcher finds the PO and goods receipt; line 3 unit price is above the PO by a few percent.
- The variance exceeds tolerance, so the item goes to a review queue with the difference highlighted.
- The buyer accepts or disputes; on accept, the draft is created with a comment.
Result: Routine invoices flow without keying, while price exceptions get a person's attention at once.
Mill certificate check on receipt
Trigger: A delivery is received for a material that requires a certificate.
- Receiving uploads a photo of the certificate.
- OCR reads heat number and key properties.
- The pipeline compares heat number with delivery batch and flags a mismatch.
- The certificate is attached to the delivery or batch record.
Result: Material with missing or mismatched paperwork is caught at the dock, not at audit time.
Keeping data on-prem
ByDesign is hosted by SAP, but invoices, certificates and drawings often carry pricing, contract terms or controlled technical data you may not want in a public OCR API. Run OCR and extraction in your own network with open-source OCR and an open-weight vision-language model on local GPUs, and store source documents in your own storage. Only the resulting structured drafts and the attachment need to go to ByDesign through its services.
Security checklist
- Encrypt documents at rest and restrict access to the review UI by role.
- Keep the posting identity limited to draft creation; payment release must stay under ByDesign approval rules.
- Check bank detail changes on invoices against the master data and flag differences as potential fraud.
- Segregate duties: the person who approves a corrected OCR result should not also release payment.
- Keep an immutable record of original file, extracted values, edits and who approved.
Ways to get it done
Cloud invoice-capture service
Effort: 2 to 6 weeks including integration.
Fast for standard invoices if data may leave your environment; confirm it can post to ByDesign.
Open-source OCR with in-house review UI
Effort: 4 to 10 weeks.
Cost-effective when you have developers and a limited set of document types.
Private vision model pipeline via Netray or a partner
Effort: 4 to 8 weeks for first document types.
Best for mixed layouts, scanned certificates and drawings, with data staying on your infrastructure.
Pitfalls
- !Trusting extracted totals leads to posting wrong amounts; always recompute and reconcile.
- !Credit notes and invoices with negative lines are routinely mis-signed; test them explicitly.
- !Tax codes and withholding vary by country and supplier; map them deliberately rather than inferring.
- !Phone photos of packing slips are skewed and shadowed; add capture guidance and pre-processing.
- !A threshold set too aggressively floods reviewers, too loosely lets errors through; tune it per document type with real data.
Frequently asked questions
Does ByDesign include OCR for supplier invoices?
Some releases or add-ons may offer invoice capture features, but this changes by version and licence. Ask your SAP partner what is available in your tenant. If it covers your documents and approval rules, use it before building your own.
How accurate is OCR for invoices?
For clean digital PDFs with consistent layouts, extraction of key header fields is usually high. Scanned, skewed or handwritten documents are noticeably worse. Judge it on your own samples and track field-level accuracy rather than relying on a vendor headline.
Can OCR create the invoice automatically without review?
It can, for cases that pass strict checks: known supplier, matching PO and goods receipt, totals reconcile, no duplicates. Most companies keep review for exceptions and for new suppliers until trust is established.
How are drawings handled?
OCR can read title block data such as part number and revision, but it does not understand the drawing. Use it to index and link drawings to documents, and do not treat extracted values as engineering truth without review.
Where should the original document be stored?
Attach it to the ByDesign document so users can see it, and keep the master copy in your document store with retention rules. Storing in both is common, with a document ID linking them.
Want Document OCR working on your SAP Business ByDesign?
A free 30 minute assessment: we map the integration path for your version and setup, and where it should run.
Get a free assessment