Migrating From File Servers and Shared Drives

For every SharePoint migration we run, we run three file-server migrations. The Windows file server (often: an old Server 2012, mapped as S:\ on every user's PC, full of final.docx, final_v2.docx, final_FINAL.docx) is still the dominant document store in Kenyan enterprises.

This is the playbook.

The starting state

A typical Kenyan file server has:

200-500 GB across 50K-300K files
Deeply nested folders (8-12 levels not uncommon)
Broken permissions (Everyone has Modify because somebody forgot)
40-60% duplicate content
10-30% files in formats that haven't been opened in 5+ years (.doc, .xls, .ppt)
A reasonable percentage of personal stuff users dumped there

You will not migrate all of this. You should not migrate all of this.

The four-phase approach (16 weeks)

Phase 1 — Shadow indexing (weeks 1-2)

Install the Papyrus Desktop Agent on the file server itself. Point it at the share root in passive mode (no upload yet, just inventory).

Output: a catalogue of what's there, by:

Type
Age (last-touch date)
Size
Permission complexity
Duplicates

This catalogue tells you what's worth migrating.

Phase 2 — New-file capture (weeks 3-4)

Switch the Desktop Agent to active mode. Any new file placed in the share auto-uploads to Papyrus. Existing files don't move.

This is the most important phase. New documents stop accumulating in the old share. The migration backlog stops growing.

User-visible change: zero. They keep saving to the share. Papyrus ingests in the background.

Phase 3 — Hot file migration (weeks 5-12)

Move files touched in the last 18 months to Papyrus folders, preserving structure (initially).

This is roughly 15-25% of file count but 80% of daily access patterns. After this phase, “most” of what users care about is in Papyrus.

Set up redirect notices on the migrated folders (“This folder has moved to Papyrus — search there”).

Phase 4 — Cold archive and decommission (weeks 13-16)

Decide what to do with the cold tail:

Migrate to Papyrus as cold storage: low-priority ingest, full classification but no notifications
Archive externally: dump to a cheap blob store; keep the catalogue in Papyrus pointing to the external location
Discard: for content beyond retention with no business value

By week 16, the file server is read-only or off.

What people will fight you on

"But I know where everything is in S:"

The classic protest. Counter with: search in Papyrus is faster than navigating S:, especially when the user doesn't know exactly where they put it. Demo it.

“My team's structure is sacred”

Allow it to be preserved as Papyrus folders for the first 90 days. After that, when users have built confidence with search, the folder structure starts to feel unnecessary. They'll abandon it themselves.

"I have personal files on S:"

Tell them to extract those before migration. Set a deadline. Then run an exclusion filter for known personal patterns (private photos, CVs, tax returns) — those don't migrate.

"What about [system X] that reads from S:?"

Identify these in Phase 1. Most can be re-pointed to Papyrus via the API. Some legacy systems may need a fileshare facade — Papyrus can expose a virtual file path for backward compatibility on Enterprise plans.

What goes wrong

Mode collapse

You set up Phase 2 (new-file capture), then never proceed. The file server stays as a “warm” store while Papyrus is the “active” store. Six months later, you have two systems instead of one.

The fix: lock the calendar dates for Phases 3 and 4 before starting Phase 2.

Permission inflation

You bring file-server permissions over wholesale. Papyrus is now “Everyone can see everything”, same as the file server, but now with audit logs. Auditors find it.

The fix: take this opportunity to reduce permission scope. Default to department-only access; grant more on request.

Behaviour persistence

Users save to S:\ even after migration because muscle memory. Auto-redirect S:\ to a watched folder in Phase 4. The Desktop Agent captures the saves silently.

Success criteria

After 16 weeks, you should be able to say:

“All new documents land in Papyrus” — true
“All documents touched in the last 18 months are in Papyrus” — true
“The file server is read-only” — true
“Users find documents at least as fast as before” — true (typically faster)
“Audit log shows who accessed what when” — true (the file server never could)