Most EDM (Electronic Document Management) projects fail for the same reason: the company copies the file server folder structure into SharePoint and calls it a migration. The result is the same chaos, now with a licence cost on top. Real document management starts when you stop describing a document by where it sits and start describing it by what it is.
1. Map document types before touching the tool
Before creating a single library, list the document types that actually circulate: contract, invoice, standard operating procedure, technical report, internal policy, minutes. For each one, record who creates it, who approves it, how long it must be kept and how confidential it is.
This exercise usually reveals that a mid-sized company works with 15 to 25 real types - not the 4,000 folders sitting on disk. Those types are the backbone of everything else.
2. Fewer libraries, more metadata
The classic mistake is one library per department, each replicating the old maze of subfolders. The alternative is a small number of large libraries organised by lifecycle and permission, using metadata columns for everything that used to be a folder.
- Folder: one fixed path, a single way to see the archive.
- Metadata: filters by team, status, year and owner - all at once.
- Rule of thumb: if the information helps find the document later, it is a column, not a folder.
Use the Term Store for vocabularies repeated across libraries: business units, cost centres, categories. That way "Legal" is always the same term rather than seven hand-typed spellings.
3. Content types: the contract between IT and the business
A content type binds required columns, file template, retention policy and approval flow into a single object. Published from the content type hub, it propagates to every site that consumes it.
In practice this means "Contract" always carries the same columns - counterparty, term, value, owner - no matter which site it was created in. That is what makes a consolidated contract report possible six months later.
4. Versioning and check-out without blocking the team
Enable major and minor versioning on libraries holding living documents (procedures, policies, manuals). Readers see only the latest published version while authors work through drafts without publishing noise.
Mandatory check-out only makes sense for low-volume, high-criticality libraries. In a daily-use library it becomes a bottleneck: someone goes on holiday with a file locked and the process stops.
5. Retention and disposal: the part nobody plans
Document management is not only about keeping - it is about knowing when to dispose. A document kept beyond its legal term is a liability, not an asset: it widens the audit surface and complicates data subject requests.
Configure retention labels aligned to the company retention schedule and apply them automatically by content type. Disposal must be a designed process with a disposition review for sensitive archives - not a manual clean-up triggered when storage runs out.
6. Adoption: the project ends when habits change
Make the new path easier than the old one: default column values, ready-made views per team, the library pinned in the Teams channel people already use - and the old file server set to read-only on an announced date.
If a user must remember seven fields to save a file, they will save it to the desktop. Every required field has to justify its own cost.
Go-live checklist
- Document types mapped and validated with the business.
- Libraries defined by permission and lifecycle, not by org chart.
- Content types published from the central hub.
- Term Store loaded with corporate vocabularies.
- Versioning configured per archive criticality.
- Retention labels applied by content type.
- Views ready and library pinned in Teams.
- Legacy file server read-only with a communicated cut-off date.
With these six stages closed, document management stops being a repository and becomes process infrastructure - which is exactly what makes the next step possible: automating document approval.