Unstructured data: The missing piece in SAP transformation

Key takeaways

• Around 80% of enterprise data is unstructured, and most transformation tools aren’t built to handle it.
• Left unmanaged, it creates real risk: manual reworking, compliance exposure, and shaky day 1 readiness.
• AI can make this data usable at scale, but transformation expertise is still what decides what actually happens to it.

What structured data alone doesn’t capture

Transformation projects are heavily geared toward structured ERP data. Meanwhile, a large share of business-critical information sits outside these structures – in formats that require a different approach to assess and manage.

Common examples include:

• Contracts and NDAs
• Employee records
• Work instructions and process documentation
• Sales documents
• Emails
• IT documentation and support tickets
• PDFs, presentations, and spreadsheets stored outside the ERP system

Most organizations have this data spread across dozens of systems: SharePoint sites, shared drives, inboxes, and sometimes paper records that were never digitized at all.  These repositories were built for day-to-day business operations, not for future carve-outs, mergers, or system transformations.

That’s precisely why this issue becomes critical during projects like these. An M&A is the one moment when a company’s information is funneled through a single point, often on a tight deadline. Teams suddenly need to work out what information belongs to the entity being separated, what should be migrated, and what can be left behind – without the luxury of time to review it document by document.

According to Gartner, unstructured data such as documents and multimedia files makes up around 80% of enterprise data. Most of it has never been cataloged, let alone scoped for a transformation.

Where unstructured data creates risk

Unstructured data that isn’t properly managed doesn’t just sit there quietly. It creates risks that tend to surface later.

These risks compound quietly. A single missing bill of materials or an outdated legal document can go unnoticed for months. Then, when it’s needed, no one can locate it or verify that it’s still current.

The core challenge isn’t volume, either. It’s knowing which information is relevant, business-critical, or sensitive, and which isn’t. Reviewing everything by hand doesn’t scale. Ignoring it isn’t an option.

How AI helps you work with unstructured data at scale

AI is starting to change how organizations approach this issue. Instead of relying on file names or keywords, AI can interpret what a document actually contains and why it matters.

At a high level, AI-powered capabilities can help teams:

• Scan and analyze large volumes of documents
• Classify content based on meaning and context, not just labels
• Identify business-relevant or sensitive information
• Apply consistent rules across large data sets
• Reduce repetitive manual review
• Support more informed decisions about what to retain or migrate

Gartner frames this as a five-step discipline: Discover and catalog the data, preprocess and analyze it, tag and classify it, connect it to structured systems, and govern it with clear policies throughout. The result is unstructured data moving from a manual, reactive process toward something systematic, scalable, and context-aware.

AI alone isn’t enough, though. For the analysis to be useful in a real transformation, it has to connect to the business context and the specific goals of the migration or carve-out, not just produce a tidy list of tagged documents.

How AI brings unstructured data into the transformation

AI can tell you what a document is. It can’t always tell you what to do with it.

Take a contract, for example. AI can identify it, classify it, and flag it as relevant.

What AI may still need additional context to determine:

• Which legal entity it belongs to
• Whether it’s still active
• Whether it should be migrated, retained, or archived
• Where it belongs in the target environment

Those are transformation decisions, shaped by the same business-driven scoping and validation that any well-run migration depends on. AI doesn’t replace that judgment. It gives the people making those calls a far better starting point than a shared drive full of unlabeled files.

SNP’s approach to unstructured data

SNP is extending its Kyano platform in this direction. An early capability, Unstructured Data Processing, applies AI-powered analysis to unstructured data as part of the broader transformation process.

In practice, SNP is already applying this approach to unstructured data. Here are two examples:

• A multi-year migration where critical plant documentation turned out to exist only as scattered PDFs and paper records, known only to one person on-site.
• Decades of technical documentation from multiple acquisitions, sitting in inconsistent formats, effectively unusable without a major manual review effort.

The kind of gap that goes unnoticed until it’s urgent.

The analysis is designed to take place within the customer’s own environment, so the data doesn’t need to be transferred to an external system. It’s early, but the direction is clear: Unstructured data is becoming a standard part of how transformations are scoped, not an afterthought discovered mid-project.

Looking ahead

Future transformation projects won’t treat structured and unstructured data as separate problems. They’ll need to manage both together from day one of planning.

Moving ERP data cleanly has always mattered. Increasingly, so does understanding and managing the broader information landscape around it: The contracts, the records, and the documents that don’t live in SAP but still decide whether a transformation actually succeeds.