Invoice capture in SAP Business One works best when you treat OCR as the first of several quality gates. For reference, Ardent Partners puts the average manual invoice at $12.88 to process and 17.4 days to clear, and every blurry scan sends an invoice back into that manual lane.
In this post, you’ll find the scan settings that matter, a six-stage capture workflow for B1, the quality checks to run before anything posts, and a checklist your AP team can use every day.
What is invoice capture and OCR in SAP Business One?
Invoice capture in SAP Business One is the process of turning supplier invoices (paper, PDF or email) into structured A/P invoice data. OCR reads the text from the image, then capture software maps vendor, invoice number, date, tax and line items to B1 master data so the invoice can be matched, approved and posted.
The distinction matters because OCR only reads characters; it does not know your vendors, your G/L accounts or your open purchase orders. Everything that makes the data usable in B1 happens in the validation layer after recognition.
How native invoice OCR works in SAP Business One
SAP introduced the Document Information Extraction service for SAP Business One in version 10.0 FP 2102. It reads supplier invoices and brings them into B1 as structured A/P invoice drafts.
Under the hood, B1’s Electronic Document Service sends PDF and JPG files to the extraction service on SAP Business Technology Platform, and the Electronic Documents Import Wizard turns the JSON response into drafts, as SAP’s integration overview describes.
The service is licensed separately from SAP Business One, and it needs BTP credentials, polling settings and mapping rules before it produces anything useful. In practice, native OCR is a configuration project, not a toggle.
Tax is a known soft spot. SAP Note 3577712 explains that drafts created through the import wizard use the default tax code the system identifies, so someone still has to check tax before posting.
What SAP Business One handles, and what it leaves to you
B1 is a strong system of record for posted AP data. It handles these natively:
- Vendor and item master data
- Purchase orders and Goods Receipt POs (GRPOs)
- A/P invoice posting and payment runs
- Blocking duplicated vendor reference numbers, if you switch it on in Document Settings
What it leaves to a person or a capture layer:
- Pulling invoices from every inbox, scanner and supplier portal into one queue
- Extracting line items reliably from varied supplier layouts
- Matching invoices to POs and GRPOs before posting, with tolerances
- Routing approvals by amount, vendor, department or entity, with reminders
- Keeping a single audit trail of who changed what before the invoice reached B1
That second list is why most B1 teams end up evaluating AP automation tools built for SAP Business One rather than relying on scanning alone.
Scanning best practices for accurate invoice OCR
Recognition quality is decided at the scanner. No extraction model can read detail the image never captured.
Scan at 300 dpi in grayscale, with matching resolution on both axes
ABBYY’s scanning guidance recommends 300 dpi in grayscale for printed documents, warns that anything under 150 dpi hurts recognition, and notes that going above 600 dpi slows OCR without meaningful gains. ABBYY also requires horizontal and vertical resolution to match, so check that your multifunction printer is not using a fax-style 204 x 96 setting.
Google’s Document AI documentation sets 200 dpi as the floor and says 300 dpi or higher generally produces the best results. Set 300 dpi as your default profile and lock it so nobody changes it to save disk space.
Give dense line items extra resolution
Small print is where line-item extraction breaks first. Microsoft’s Document Intelligence documentation puts the minimum text height at 12 pixels on a 1024 x 768 image, roughly 8-point text at 150 dpi.
Freight bills, utility invoices and distributor statements often print line items at that size or smaller. For those suppliers, a 400 dpi profile is a cheap fix.
Use lossless formats and stop the recompression
Google’s same guidance warns that lossy formats such as JPEG can lose image quality, and extraction accuracy, when file sizes are reduced. Have the scanner output PDF or TIFF, and stop staff from forwarding phone photos through chat apps that compress them.
One invoice per file, in order, right side up
Most “OCR errors” in a B1 exception queue are really batching errors. Set these rules at the scanner:
- One invoice per file: use separator sheets or the scanner’s blank-page split for batches
- Keep multi-page invoices together, with pages in order
- Remove staples and flatten folds before feeding
- Turn on auto-deskew and auto-rotate
- Scan the reverse side only when it carries content such as payment terms
Prefer the supplier’s PDF over paper
A supplier’s original PDF usually carries a text layer, so it skips image recognition errors entirely. Every vendor you move from paper to emailed PDF removes a scanning step for good.
Start with your 20 highest-volume paper suppliers and ask them to email a dedicated AP address. Keep the scanner for whatever is left.
Designing an invoice scanning workflow for SAP B1
A reliable workflow has six stages. Each stage needs a named owner and a defined output, or invoices pile up between them.
1. Intake: one address, one queue
Give suppliers a single AP email address and route scanned mail to the same queue the same day. Scattered personal inboxes are where invoices go missing before anyone captures them.
2. Capture and extraction
Extract the header (vendor, invoice number, dates, currency, totals, tax) and every line item. If you only capture headers, you cannot run line-level PO matching later, which is where pricing and quantity errors hide.
This is the stage that automated AP invoice processing replaces first, because it is the most repetitive work in the cycle.
3. Validation against SAP Business One master data
Map every extracted value to a live B1 object: business partner code, G/L account, cost center, project, branch and tax code. A value that cannot be mapped should stop the invoice, not get a default.
4. Matching to purchase orders and goods receipts
Match the invoice to its PO (two-way) or to the PO and GRPO (three-way). Set price and quantity tolerances by vendor or item group, and send anything outside them to review.
5. Approval routing
Route non-PO invoices by amount, vendor, department or entity, with automatic reminders and escalation. An approval with no named owner is the most common reason an invoice misses its due date.
6. Posting and archiving
Post the approved invoice to B1 as an A/P invoice with the original document attached. The ERP stays the system of record, and the attachment gives auditors the source without a trip to the filing cabinet.
Quality checks to run before an invoice posts to SAP B1
Yes, you can implement invoice quality checks for SAP Business One at three layers: image checks at scan time, field-level validation after OCR, and business-rule checks (duplicates, PO matching, approvals) before posting. B1’s Document Settings cover duplicate vendor references, and most other checks need an approval procedure, a stored procedure or a capture layer.
Image-level checks
- Resolution is at least 300 dpi, or 400 dpi for small-print suppliers
- The page is upright, deskewed and fully in frame
- Page count matches the invoice’s own “page x of y”
- No handwritten notes cover totals, dates or the invoice number
Field-level checks
- All required fields are present: vendor, invoice number, invoice date, currency, net, tax and gross
- Line totals plus tax equal the invoice total
- The invoice date is not in the future and not older than your policy allows
- The currency matches the business partner’s currency in B1
- The tax code matches the vendor’s usual tax treatment
- Bank details on the invoice match the vendor master, and changes go to a separate verification step
Duplicate checks
SAP Business One can block a duplicated vendor reference number per document type through Document Settings, as SAP’s guide to tracking reference numbers explains. Turn it on for A/P invoices with “Block Release”.
That check has limits. SAP Note 3262546 confirms a vendor reference from a cancelled A/P invoice can be reused, and the block only works if the reference is keyed the same way every time.
Add a fuzzy check on vendor, amount and date to catch the same invoice entered as “INV-1001” and “1001”. Duplicate and bank-detail checks are also your first line of fraud mitigation and compliance in AP.
Match and approval checks
- Every PO invoice matches its PO, and its GRPO where goods were received
- Price and quantity variances sit inside tolerance, or carry a reviewer’s note
- Partial deliveries are matched against the right GRPO lines
- Non-PO invoices have an approver assigned by rule, not by memory
- Nothing posts without a recorded approval
Confidence thresholds and exception queues
Most capture tools score how sure they are about each field. Route any low-confidence field to a person instead of posting it, and track which vendors trigger the most reviews.
Work the exception queue daily, by reason code. Fixing the cause (a new layout, a missing BP code, a bad scan profile) is worth more than fixing the invoice.
Common invoice OCR problems in SAP B1 and how to fix them
| Problem | Likely cause | Fix |
|---|---|---|
| Characters misread (8 for B, 0 for O) | Scan under 200 dpi, or a recompressed JPEG | Rescan at 300 dpi grayscale and output PDF or TIFF |
| Line items missing or merged | Small font and dense tables | Use a 400 dpi profile for that vendor and a capture tool that extracts lines, not only headers |
| Wrong business partner selected | Name variants or duplicate BP records | Match on tax ID or bank details and clean up duplicate BP codes |
| Wrong tax code on the A/P draft | Native import applies the default tax code | Set a tax rule per vendor and review tax before posting |
| Duplicate invoice posted | Reference keyed differently, or reused from a cancelled document | Block duplicates in Document Settings and add a fuzzy vendor-amount-date check |
| One invoice split into two documents | Batch split on the wrong page | Use separator sheets and keep multi-page invoices together |
| Invoice stuck before posting | No owner, no reminder | Assign approvers by rule, with automatic reminders and escalation |
Metrics that show your capture process is working
Track a small set of numbers monthly, and review them alongside the exception queue:
- First-pass rate: share of invoices that post without a manual correction
- Exception rate by reason: image quality, missing PO, price variance, unknown vendor
- Corrections per vendor: shows which supplier layouts need attention
- Cycle time: receipt to posting in B1, and receipt to approval
- Duplicates caught before posting: proof that the checks are doing their job
- Cost per invoice: model it with the AP automation ROI calculator to see what a higher first-pass rate is worth
The SAP B1 invoice capture and OCR checklist
Print this, or paste it into your AP procedures. Review it monthly against your exception report.
Scanner and image setup
- ☐ Default scan profile locked at 300 dpi, grayscale, equal horizontal and vertical resolution
- ☐ 400 dpi profile available for small-print suppliers
- ☐ Output set to PDF or TIFF, never recompressed JPEG
- ☐ Auto-deskew, auto-rotate and blank-page removal turned on
- ☐ Separator sheets or batch-split rules in use
Intake
- ☐ One dedicated AP email address published to all suppliers
- ☐ Paper mail scanned into the same queue on the day it arrives
- ☐ Top paper suppliers asked to switch to emailed PDF
- ☐ No invoices processed from personal inboxes
Extraction and validation
- ☐ Header and line items captured for every invoice
- ☐ Every vendor mapped to an active B1 business partner code
- ☐ G/L account, cost center, project and branch validated against B1
- ☐ Totals, tax and currency checks run automatically
- ☐ Tax code reviewed before posting
- ☐ Low-confidence fields routed to review, not posted
Matching and approval
- ☐ Two-way or three-way match against PO and GRPO
- ☐ Price and quantity tolerances set by vendor or item group
- ☐ Approval rules defined for non-PO invoices
- ☐ Reminders and escalation active for overdue approvals
- ☐ Bank-detail changes verified outside the invoice channel
Posting and audit
- ☐ Duplicate vendor reference blocked for A/P invoices in Document Settings
- ☐ Fuzzy duplicate check on vendor, amount and date
- ☐ Original invoice attached to the posted A/P invoice
- ☐ Full audit trail of edits and approvals retained
Monthly review
- ☐ First-pass rate and exception reasons reviewed
- ☐ Worst five vendors by corrections contacted or reconfigured
- ☐ Scan profiles spot-checked on a sample of ten invoices
- ☐ Cycle time and cost per invoice compared with last month
Case study snapshot: how Bolla Market moved high-volume AP off paper
Bolla Market (Bolla Oil Corp.) runs one of the highest-volume AP operations with DOKKA’s AP automation platform. The before-and-after shows what changes when capture, approval and posting live in one workflow.
- Company: 350+ gas stations, car washes and convenience stores across New York and New Jersey, 2,000+ employees
- AP scope: about 10 companies in one AP operation
- ERP: SAP, with a native DOKKA integration
- Before: paper invoices in folders, approvals signed by hand, then scanned so an offshore team could key them into SAP; nobody could see where an invoice stood, and documents went missing
- After: invoices arrive by dedicated email address, by forwarding from a central AP inbox, or as scanned paper; key fields are captured automatically and extraction improves as each vendor’s format is learned; fully approved invoices are published to SAP for payment scheduling
- Results: up to 1,000 invoices a day across about 10 companies at roughly one minute each, a named approver on every invoice, and a daily lifecycle and aging report for the CFO
“The main thing I love about it is that it’s able to track where things are.” Keith Sherman, Accounts Payable Manager, Bolla Market
Notice what changed: the scanner stayed, the re-keying went. Read the full Bolla Market case study for the rollout details.
Moving beyond template OCR in SAP Business One
Template OCR reads fields; it does not decide whether an invoice is ready to post. DOKKA’s agentic AP automation for SAP Business One runs the checks in this guide as agents that execute the work while your team reviews it.
- Invoice Processing Agent: extracts headers and line items, learns coding behaviour over time, and flags missing or inconsistent data
- PO Matching Agent: runs two-way and three-way matching against B1 POs and GRPOs and surfaces price and quantity discrepancies
- Approval Agent: suggests approvers, sends reminders and escalates overdue approvals
- Document Agent: detects duplicate invoices and keeps a complete audit trail
DOKKA pulls vendors, G/L accounts, cost centers, projects, open POs and GRPOs from B1, then pushes back A/P invoices, GRPOs, journal entries and the original documents. SAP Business One stays the system of record, and payment runs stay in B1.
Teams typically go live in 2 to 4 weeks. You can see how the four agents work together on the DOKKA AP platform page, and check which ERPs are supported on the integrations page.
Frequently asked questions
What is the best scan resolution for invoice OCR?
300 dpi in grayscale is the standard recommendation for printed invoices. Go to 400 dpi for very small print, and avoid going below 200 dpi or above 600 dpi, where accuracy drops or processing slows with no real gain.
Does SAP Business One have built-in invoice OCR?
Yes, through the Document Information Extraction service on SAP Business Technology Platform, available since version 10.0 FP 2102. It creates A/P invoice drafts from PDFs and JPGs, but it is licensed separately and needs configuration, mapping and review steps before invoices post.
Can SAP Business One block duplicate supplier invoices?
Yes, partly. Document Settings can block a duplicated vendor reference number on A/P invoices, but it misses references keyed differently and allows reuse from cancelled documents, so add a vendor, amount and date check on top.
How accurate is invoice OCR?
Accuracy depends mostly on image quality, font size and layout variety, so a single headline figure tells you little. Measure field-level accuracy on your own top vendors, and track the first-pass rate as your real benchmark.
Should we scan paper invoices or ask suppliers for PDFs?
Ask for PDFs wherever you can. An emailed PDF usually carries a text layer, so it avoids scanning errors entirely and arrives faster than paper.
Which quality checks should run before an A/P invoice posts in SAP B1?
At minimum: required fields present, totals and tax reconciled, vendor and G/L mapped to live B1 records, a duplicate check, a PO or GRPO match where one exists, and a recorded approval. Anything that fails goes to an exception queue, not into B1.
How long does it take to automate invoice capture for SAP Business One?
Native Document Information Extraction depends on your partner’s BTP setup and mapping work. DOKKA’s native SAP Business One integration typically goes live in 2 to 4 weeks.
Conclusion: get the image right, then let the checks do the work
Good invoice capture in SAP Business One starts with a disciplined scan profile and ends with checks that stop bad data before it posts. The scanner sets the ceiling on accuracy; your validation rules decide what reaches the ledger.
Use the checklist above as your baseline, measure first-pass rate monthly, and fix causes rather than invoices. If you want to see capture, matching and approvals running against your own B1 data, book a DOKKA demo.