SAP-DMS integration: avoid 5 costly mistakes
The 5 mistakes that derail SAP-DMS projects — and how to avoid them: single source of truth, ArchiveLink, link not sync, event-driven flows, master data.
Kenan Trgic8 min read
A SAP and DMS integration connects the SAP ERP system with a Document Management System so that business documents — invoices, delivery notes, contracts, drawings — are linked bidirectionally to the corresponding SAP transactions and stored once, in a governed archive. eelik d.o.o. designs and builds these integrations for manufacturing and services companies in the DACH region, using ArchiveLink, REST API and event-driven patterns. Most integration failures we are asked to fix are not technical bugs but design decisions that were made — or skipped — in the first two weeks of the project.
Why is the SAP–DMS interface so often the weak point?
SAP and a DMS are both systems of record, each with strong opinions about how documents and metadata should be stored. When the integration is treated as a back-of-the-project IT task instead of an architectural decision, organisations end up with documents in both systems, conflicting metadata, and no single answer to the question "where is the authoritative copy?". The five mistakes below account for the majority of the integration projects we are called in to remediate.
Mistake 1 — Not defining the single source of truth up front
The most expensive mistake is the simplest to describe: nobody decided whether SAP or the DMS owns the document. Six months in, the same invoice exists in both systems with slightly different metadata, and reconciliation work consumes the productivity gain the project was supposed to deliver.
How it shows up: duplicate PDFs in SAP DMS tables and in the external DMS; user confusion over which system to search; audit findings about inconsistent retention dates.
Business impact: failed audits, doubled storage cost, and erosion of user trust in both systems.
Mitigation: decide per document class. A defensible default for DACH organisations is that SAP owns the transactional data (invoice header, posting, accounting view) and the DMS owns the document object (PDF, archive, retention, audit). SAP holds a link, not the file. Document this decision in writing before any code is committed.
Mistake 2 — Synchronising files instead of linking them
Some teams try to "keep both systems in sync" by replicating documents between SAP attachments (GOS — Generic Object Services) and the DMS. This works for a week, then drifts.
How it shows up: version mismatches, attachments missing in one system, file size and storage bills doubling.
Business impact: storage cost grows linearly with transaction volume, and you lose the legal certainty of a single archived original.
Mitigation: use ArchiveLink or a content repository interface so that SAP stores only a reference (DocID, content repository ID) and the file itself lives once in the DMS. The DMS becomes the content store; SAP becomes the metadata and link store. SAP's own ArchiveLink documentation describes the interface in detail.
Mistake 3 — Ignoring ArchiveLink and reinventing the wheel
ArchiveLink is the standard SAP interface for archiving and linking documents to business objects. It has existed since the 1990s, is supported by every certified DMS, and handles inbound (incoming invoices), outbound (printed invoices, delivery notes) and late-archiving scenarios.
How it shows up: custom ABAP code that calls a REST API directly, bypassing GOS, with no link visible from MIRO, FB03 or VA03 transactions.
Business impact: end users cannot find documents from their normal SAP screens, training cost increases, and you have built a custom integration that the next consultant will not understand.
Mitigation: use the ArchiveLink interface for everything that has a business object reference in SAP (FI invoices, SD orders, MM deliveries, HR personnel records). Reserve REST or OData integrations for the cases where ArchiveLink is genuinely not suitable, such as documents that live outside SAP's object model.
Mistake 4 — No change-data-capture for near real-time updates
Most integrations start as nightly batch jobs. That is acceptable for archive transfer but breaks down the moment finance wants to see scanned invoices in SAP within minutes, or sales needs the signed contract visible immediately after DocuSign returns it.
How it shows up: users complain that "the document isn't there yet"; AP clerks re-scan invoices because the batch has not run.
Business impact: the integration is rebuilt twice — once with batch, once with events — at roughly double the original cost.
Mitigation: plan event-driven flows from day one. SAP events (SAP Change Documents via CDHDR/CDPOS, BAdIs, or SAP Event Mesh) push notifications to the DMS; the DMS pushes events back via REST webhooks when a document is captured or approved. Batch jobs remain only for bulk historical migration and nightly reconciliation.
Mistake 5 — No governance for master data
Vendors, cost centres, plants, document types — these are the metadata fields the integration depends on. If they are maintained independently in SAP and in the DMS, the link breaks the first time a vendor number is renumbered or a document type is renamed.
How it shows up: documents land under "unknown vendor"; reports show drops in classification accuracy after master data clean-ups.
Business impact: silent data quality degradation that is usually only discovered during the annual audit.
Mitigation: SAP is the master for business master data (vendors, customers, cost centres, GL accounts). The DMS subscribes to these via daily delta or event push. Document types and retention classes can be mastered in the DMS but mirrored read-only into SAP. Establish a written master-data governance document and review it quarterly.
How should a clean SAP–DMS integration look?
| Layer | Responsibility | Pattern |
|---|---|---|
| Document storage | DMS only | Single archive, retention rules |
| Business object link | SAP holds DocID + ContentRepID | ArchiveLink |
| Metadata master | SAP for business data, DMS for document classes | Event-driven replication |
| Capture (incoming invoices) | DMS / capture platform | OCR + NLP, posted to SAP via BAPI/IDoc |
| Output (printed documents) | SAP generates, DMS archives | Outbound ArchiveLink |
| Search | Both systems, federated | SAP GOS links to DMS view |
| Audit | DMS owns the audit trail of the file; SAP owns the audit of the transaction | Tamper-evident logs in both |
This layered model removes the five mistakes by construction: the source of truth is explicit, files are not duplicated, ArchiveLink is the default, events drive freshness, and master data has a clear owner.
What about S/4HANA migration?
The move to S/4HANA is a good moment to revisit SAP–DMS integration. Many organisations carry forward legacy GOS attachments that should have been archived years ago. A pre-migration archive cleanup, combined with an ArchiveLink-based external DMS, reduces the size of the migration and improves S/4 performance afterwards. eelik d.o.o. runs these archive consolidation projects in parallel with S/4 migration partners.
How long does a typical integration take?
For a mid-sized SAP estate (one production system, FI/MM/SD modules, 200–500 named users):
- Discovery and architecture: 3–4 weeks
- Inbound invoice flow with ArchiveLink and capture: 6–8 weeks
- Outbound printed documents: 3–4 weeks
- Master data and event flows: 4–6 weeks in parallel
- Go-live and stabilisation: 4 weeks
Total: roughly four to five months for the first wave. See the system integrations service and the document management service for the delivery model.
Frequently Asked Questions
Do we still need ArchiveLink in S/4HANA?
Yes. ArchiveLink remains the standard interface in S/4HANA, both on-premises and in private cloud. S/4HANA Cloud Public Edition replaces the classic customer-administered Content Server with SAP Document Management Service, so the archive lives in the SAP-managed service rather than in a custom ArchiveLink content repository. Certified DMS partners integrate through this service. On-premises and Private Cloud editions continue to support classic ArchiveLink without restriction.
Can we use REST API instead of ArchiveLink?
For documents that have no SAP business object (general correspondence, generic archives) a REST or OData integration is reasonable. For anything attached to an FI document, sales order, delivery or personnel record, ArchiveLink remains the lower-risk choice because it integrates with the standard SAP UI and authorisation model.
What if our SAP system is hosted by a third party?
The integration model does not change. The DMS connects to SAP via secure VPN or AWS/Azure PrivateLink, and ArchiveLink calls run over the same channel. Confirm with the hosting partner that custom RFC destinations and ArchiveLink content repositories can be configured.
How do we avoid a failed project?
Spend the first two weeks on the architecture decisions in this article — single source of truth, link-not-sync, ArchiveLink-first, event-driven, master-data governance — and write them down. Most failed integrations skip this and pay for it in months six through twelve. Contact eelik d.o.o. for an integration assessment.
Related service
Services
About the author

Founder & CEO
IT consultant with over 12 years of experience in enterprise content management, system integrations, and digital transformation. Specialized in DMS and ECM implementations across Central Europe.
LinkedIn