What moves, what doesn't (the planning table)
| Artifact | Migratability |
|---|---|
| Users/identities | Well-trodden (new objects in target; identity cutover discipline) |
| Files (SPO/OD incl. channel files) | Good — mature tooling, metadata largely preserved |
| Team/channel STRUCTURE | Recreatable by tooling (names, membership, settings) |
| Channel messages | The hard wall: no supported true-fidelity migration; third-party tools re-post as imports (altered authorship/timestamps fidelity) or produce archives |
| 1:1/group chats | Practically: archive for compliance, start fresh |
| Meetings (future) | Rebooked/re-created; recordings move as files |
| Phone numbers | A porting project per country (numbers module), timed to cutover |
| Policies/config | Re-implemented — and this is the CHANCE to rationalise, not copy debt |
Strategy follows the table: archive history, migrate files, recreate structure, start conversations fresh — with eDiscovery-grade archives of the source tenant satisfying the records obligation that 'we migrated chat' never truly does.
The three-phase playbook
- Coexistence (weeks-months): cross-tenant access opened (B2B/shared channels for joint work), MTO if the interim is long, directory sync decisions made ONCE (avoid the double-B2B-object mess).
- Cutover (a weekend, per wave): identity moves, DNS/domain moves (domains detach from source and attach to target with real propagation time), files land, numbers port, day-one policies pre-staged in target (drift-scanned baseline — the serv365 pattern: snapshot source config, diff target, prove parity where intended).
- Decommission: source tenant to read-only/archival posture with holds preserved for the retention horizon — deleting the source early is how litigation finds you unprepared.
What to watch (proofs)
- The inventory that plans the work: source-tenant scan exports — teams/ channels/members, policy census (serv365's CSV is literally this), numbers, app usage — sized and owned per item.
- Migration-tool fidelity claims: a PILOT team migrated and inspected — timestamps, authorship, files-links integrity — before the claim enters the plan.
- Cutover-day burn-down: per-wave checklists with verification steps (sign-in, chat, file open, PSTN call per persona) — the wave isn't done when the tool finishes; it's done when the persona tests pass.
- Post-cutover config parity: target policy scan vs intended baseline — drift report empty = policies actually landed.