About this blog and product
why scan2datev is different?
Welcome. This blog is for everyone who uses DATEV daily and wants to understand how accounting processes can stop being manual.
What is scan2datev?
scan2datev is a microservice with a single job: take a document — PDF, scan, or photo — and return a finished, validated DATEV posting batch. No manual corrections. No import errors. Load it into DATEV and you're done.
That sounds simple. But in the DATEV ecosystem, nobody solved this for years.
The problem everyone ignored
There are dozens of tools built around DATEV. They fetch invoices from portals, digitize paper, forward emails. They all solve the same problem — moving documents from A to B.
None of them actually post.
The tax advisor ends up with a folder of PDFs. Someone has to open each one, read the numbers, assign the contra account, set the VAT key, verify the tax amount, and manually enter everything into DATEV. Or they hope that DATEV's own document recognition does the job — with a real-world hit rate somewhere between 40 and 80 percent.
scan2datev closes that gap.
What scan2datev actually does
When a document arrives in scan2datev, this is what happens:
① Multi-vector document recognition
Every document passes through three independent recognition layers. The first attempts native text extraction — when the document carries an embedded text layer, this path is fast and lossless. If the document is a scan or photo with no readable layer, a dedicated vision-based OCR engine takes over and reconstructs the full text from the image. The output of both paths feeds into a third layer: a structured extraction model that operates against a strict field schema — amounts, dates, VAT identifiers, IBAN, supplier — each slot validated on write. No free-text interpretation, no guesswork. Each layer works independently; no single point of failure can propagate to the output.
② Classification: contra account assigned automatically
366 vendor keywords and 158 description keywords map every supplier to the correct contra account — automatically for both SKR 03 and SKR 04.
③ Validation before export
128 automated checks run before a single line is written to the CSV:
- Net + VAT = Gross (tolerance 0.02 EUR)
- IBAN MOD97 verification
- VAT registration number format (DE + 9 digits)
- Document date plausibility
- Duplicate detection: the same invoice uploaded twice results in exactly one posting
④ DATEV EXTF v700 — spec-compliant
Exactly 54 columns. cp1252 encoding. CRLF line endings. Tax keys 8 (19%), 9 (7%), 94 (Reverse Charge). Credit notes with the correct debit/credit indicator. International amounts (CHF, USD, AT formats) correctly normalized.
The result: a CSV file that DATEV imports without a single complaint.
The difference in one sentence
Other tools are the postman. scan2datev is the bookkeeper.
invoicefetcher fetches invoices. GetMyInvoices collects documents. scannerbox digitizes paper. They all deliver PDFs. None of them post.
scan2datev is the first standalone service that generates a validated, import-ready DATEV posting batch from any document type — with no human intervention required.
Who is this blog for?
- Tax advisors and accounting firms who want to understand how modern document processing works
- In-house accountants who want to automate their incoming invoice workflow
- SaaS developers who need DATEV export as a feature inside their own products
- Anyone using DATEV daily who wants to understand what actually happens inside a posting batch
We write about DATEV internals, OCR, automation, and the realities of German-language accounting — without marketing language.
scan2datev is available at scan2datev.com. Try 10 documents for free.