Master Your SharePoint Implementation Timeline: A Step-by-Step Guide
Discover how to build and customize a SharePoint implementation timeline to track key phases and ensure a smooth rollout.
SharePoint Implementation Timeline: Week-by-Week Plan, Phase Durations and Limits
Most SharePoint rollouts are quoted in weeks and finish in months, and the gap is almost never the technology. Provisioning SharePoint Online takes minutes — it is already in your Microsoft 365 tenant. The calendar is set by two slower things: deciding how content should be organised — and getting those decisions signed off — and cleaning up the file shares and old sites whose content you plan to move, because years of nested folders, duplicate files and unowned documents do not migrate themselves. Plan around those two constraints and the timelines below are realistic: 4–6 weeks for a small organisation standing SharePoint up alongside a modest file share, 8–12 weeks for a mid-size organisation migrating several departments, 12–20 weeks when the estate is large, permission-heavy or spread across business units. This plan is organised phase by phase, with a duration, deliverables and exit criteria for each — because a phase you cannot declare finished is a phase that quietly eats the next one.
Last updated: November 2026. Limits verified against Microsoft Learn's SharePoint service limits documentation as of November 2026. Written and reviewed by the RolloutReady editorial team.
The numbers that set the calendar
SharePoint Online's documented limits are not fine print — they decide what your cleanup has to fix before migration, and how the migration waves are sized. Run these checks against your existing content in Phase 1, not when a migration job fails in Phase 5.
| Limit (SharePoint Online, per Microsoft Learn) | Value | What it forces in the plan |
|---|---|---|
| URL path length | 400 characters (the full path: site, folders and file name) | A path-length scan of every file share in Phase 1; over-limit files are renamed or restructured during cleanup, not during the migration weekend |
| List view threshold | 5,000 items | Libraries and lists expected to pass 5,000 items need indexed columns and filtered views designed in Phase 2 — not after users hit the threshold error |
| Items per list or library | 30 million (documented maximum) | The ceiling is not the constraint; view design is. A 200,000-item library still needs the 5,000-item view discipline |
| Maximum file size | 250 GB per file | Rarely binding for documents, but video and design archives must be inventoried in Phase 1 so a handful of oversized files do not stall a wave |
| Storage per site | 25 TB | One giant site is technically possible and architecturally wrong — the limits allow sprawl, the information architecture must not |
| Tenant storage | 1 TB + 10 GB per licence purchased | Migration volume is a storage budget: count terabytes in Phase 1 and confirm the tenant can hold the migrated estate before Phase 5 starts |
Three consequences follow. First, the 400-character path limit is the single most common cause of migration failures from file shares: a folder structure eight levels deep with descriptive file names passes it easily, and every affected file must be renamed, moved or left behind by decision — in the cleanup, on a weekday, not mid-migration. Second, the 5,000-item view threshold shapes information architecture: it is why "one library for all company documents" fails in practice and departmental sites with their own libraries succeed. Third, SharePoint Online permissions do not receive file-share NTFS permissions automatically — access has to be re-expressed as SharePoint groups and site membership during the build. Whoever plans the timeline as if permissions "come across with the files" has a surprise scheduled for launch week.
Total duration at a glance
| Rollout | Typical scope | Total timeline | Phase that decides it |
|---|---|---|---|
| Small organisation, mostly greenfield | Under 50 staff, SharePoint new, one file share under ~500 GB, few formal permissions | 4–6 weeks | Phase 2 — information architecture decisions |
| Small/medium organisation | 50–200 staff, several departments, 0.5–3 TB on file shares, some on-premises SharePoint or shared drives per team | 8–12 weeks | Phase 5 — content cleanup and migration waves |
| Larger or multi-department | 200–500 staff, multiple business units, formal governance, staged launches | 12–20 weeks | Phase 1 — discovery, ownership and sign-offs |
| Regulated or complex estate | 500+ staff, retention/compliance requirements, custom solutions to replace, heavy permission inheritance to unwind | 20+ weeks | Approvals and content decisions outside the project team's control |
The phase durations below are given as small organisation / mid-size organisation pairs. Larger rollouts mostly stretch Phases 1, 5 and 8 rather than the build.
Phase 1 — Discovery and content audit (small org: 1 week; mid-size: 2–3 weeks)
Resist the urge to create sites. This phase produces the inventory everything else is judged against: where content lives today, how much of it there is, who owns it, and how much of it should not be migrated at all. Inventory every source: file shares, personal drives destined for OneDrive, any existing SharePoint (on-premises or forgotten trial tenants), and the shared mailboxes and Teams files already in Microsoft 365. For each source, count four numbers: total size in GB/TB, file count, the count of file paths over 400 characters, and the percentage with no identifiable owner or no modification in 24 months. Those four numbers size the cleanup effort, the migration waves and the storage budget — and the last one usually cuts the migration volume by a third or more before a single file moves.
Deliverables: a source inventory per location (size, file count, over-400-character paths, stale/unowned percentage); a keep / archive / delete classification rule agreed with content owners; a named owner for every source that will be migrated.
Exit criteria: every source has an owner and measured numbers, not estimates; the classification rule is signed off by the department heads whose content it governs; the tenant storage check (migrated volume vs 1 TB + 10 GB per licence) passes or a storage purchase is approved.
Phase 2 — Information architecture and site design (small org: 1 week; mid-size: 2 weeks)
Design sites around how the organisation works and publishes, not around the folder tree of the old file share. The two building blocks are different tools: team sites are for collaboration — a group of people creating and editing content together, typically connected to a Microsoft 365 group with its mailbox, Planner and optional Team; communication sites are for broadcasting — a few authors publishing to a wide audience (intranet news, policies, HR information), with no group attached. Organise the sites under hub sites: a hub gives its associated sites shared navigation, theming and rolled-up search, so "Intranet" as one hub with department sites associated is the pattern that scales; hubs are a flat layer, not a hierarchy to nest. Fix the vocabulary of the architecture now — site naming convention, who the hub structure contains, which content types and metadata columns (department, document type, review date) are agreed centrally — because renaming and restructuring live sites is the most disruptive change there is after launch.
Deliverables: a site map (every planned site: type, owner, purpose, hub association); a naming convention for sites, libraries and groups; the agreed metadata/content-type list; one prototype team site and one prototype communication site built to the design.
Exit criteria: the site map is signed off by department heads; the prototypes pass the "find it" test — five staff who were not in the design group can each find three sample documents in under a minute; every site on the map has a named owner and a named deputy.
Decision point — design before governance tooling: metadata, content types and hub structure are cheap to change this week and expensive in Phase 7. Do not let the build phase start "while the architecture settles" — the build inherits whatever it finds.
Phase 3 — Governance and tenant configuration (small org: 3–5 days; mid-size: 1–2 weeks, overlaps Phase 2)
Governance decided after launch is not governance; it is cleanup. Configure the tenant while the architecture is being finalised: who is allowed to create sites and Microsoft 365 groups (left at the default — everyone — an organisation gets site sprawl within a quarter), the organisation-level external sharing policy in the SharePoint admin centre (Policies → Sharing), and the site lifecycle rules: how a site is requested, who approves it, what happens to a site with no activity for 12 months. Write the permission model in plain language before anyone implements it: content is secured at site level with SharePoint groups (Owners / Members / Visitors), unique permissions below site level are the exception requiring a reason, and "everyone except external users" is never the answer for departmental content. If your organisation has retention or records requirements, name the retention policies that will apply to the new sites now — they affect library design in Phase 4.
Deliverables: a written governance plan (site creation, external sharing, lifecycle, permission model); tenant settings configured to match it; a one-page site request/approval process.
Exit criteria: site creation is restricted to the agreed group or process; external sharing is set deliberately at organisation level (not left on whatever it shipped as); the permission model is signed off by whoever owns information risk.
Phase 4 — Build and configuration (small org: 1–2 weeks; mid-size: 2–3 weeks)
Build the real sites from the signed-off site map: hub structure first, then department team sites and the intranet communication sites, each with its libraries, the agreed metadata columns, and views designed against the 5,000-item view threshold (indexed columns for the fields people filter on; default views filtered, never "show everything"). Apply the permission model: SharePoint groups populated from the Phase 1 ownership lists, external sharing per site set no wider than the organisation policy. Configure the intranet home page and hub navigation so the find-it test passes on the real structure, and set up site templates or a documented repeatable build checklist so site number fifteen is built exactly like site number one. This phase is deliberately quick — if it is not, the architecture was not finished in Phase 2.
Deliverables: all sites on the site map built with libraries, metadata and views; hub navigation live; a repeatable site-build checklist; permissions applied per the model.
Exit criteria: the find-it test passes on the built structure (five staff, three documents each, under a minute); every site's Owners group contains the named owner and deputy; no library's default view will break at 5,000 items.
Phase 5 — Content migration: planning, cleanup and waves (small org: 1–2 weeks; mid-size: 3–6 weeks)
This is the phase the whole timeline bends around, and most of it is not migration — it is cleanup. Working from the Phase 1 inventory, content owners apply the keep / archive / delete rule in the source: delete the duplicates and the truly dead, archive what must be kept but not used (to cheaper storage or a read-only archive site), and only the remainder is in scope to migrate. Expect cleanup to remove 30–50% of the volume in a typical long-lived file share; every gigabyte removed is a gigabyte that never has to be renamed, moved, permissioned or trained on. In parallel, fix the 400-character path failures in what remains: shorten deep folder chains and long file names in the source, while there is still time. Then migrate in waves by department, one department per wave, each wave run the same way: a bulk copy in the background (nights/weekends for large shares), a delta pass for changes made since, permission verification against the model, and the owner signing off counts before the source goes read-only.
Tooling, honestly: Microsoft's free SharePoint Migration Tool (SPMT) handles the standard case well — file shares and on-premises SharePoint sources, bulk jobs from CSV/JSON, and incremental follow-up runs so a wave can start early and delta-sync at cutover. Reach for a paid third-party tool (ShareGate is the one most small consultancies standardise on; AvePoint and Quest Content Matrix cover larger estates) when the migration needs its reporting, permission and metadata mapping, or version-history fidelity across complex sources. The worked example below shows the decision made with numbers.
Deliverables: per-source cleanup sign-off (before/after GB and file counts); the wave schedule (department, tool, window, owner); per-wave migration reports with source vs destination counts reconciled.
Exit criteria (per wave): destination counts match source counts or every difference is documented; a sample of files opened and checked in the destination including metadata and permissions; the department owner signs off; only then does the source go read-only.
Phase 6 — Pilot department (small org: 1 week; mid-size: 2 weeks, inside the Phase 5 wave order)
Pick one department that is representative — real document volume, a mix of confident and reluctant users, an owner who will give honest feedback — and run it as wave one, completely: cleanup, migration, training and a full working week living in SharePoint while their file share is read-only. Keep the friction log: every "I can't find…", permission surprise and old-habit workaround, timestamped, reviewed twice a week. Fix in batches — the architecture and permission fixes the pilot surfaces are applied to the template (the build checklist, the training, the comms) so every later department inherits them. The pilot ends with a recorded go/no-go for the remaining waves, not with general satisfaction.
Deliverables: the pilot friction log; fixes folded into the build checklist and training materials; a recorded go/no-go decision with the remaining wave schedule confirmed or revised.
Exit criteria: pilot users complete their normal document work in SharePoint unaided for a full week; no launch-blocking defect (permissions, findability, missing content) open for 3 consecutive days; the department owner signs the wave report.
Phase 7 — Training and communications (small org: 2–3 days of sessions across 1 week; mid-size: 1–2 weeks, per department just before its launch)
Train by job, in short sessions, timed to each department's migration wave — training delivered a month before someone's content moves is training wasted. Editors and content owners get 45–60 minutes on the tasks they actually perform: uploading and metadata, sharing links versus attaching copies, version history and restoring a previous version, and where their department's content lives in the hub. Everyone else gets 20–30 minutes on finding and reading: hub navigation, search, and following the intranet. Publish a one-page quick reference per department (your sites, your library structure, who to ask) and name a champion per department — the person colleagues ask before they open a ticket. Communications run ahead of the waves: what is changing, what is not changing (file paths will; the documents themselves will not), when their department's source goes read-only, and where the old content can still be found during transition.
Deliverables: role-based session schedule delivered per wave; a one-page quick reference per department; a named, briefed champion per department; the wave communications sent (announcement, read-only notice, launch note).
Exit criteria: every member of a launching department has attended a session or watched the recording before their wave's launch day; the champion is named in the launch note; support routes (champion → owner → IT) are published.
Phase 8 — Launch in waves (small org: 1 week; mid-size: 2–3 weeks)
Launch is the per-department cutover, repeated: migration delta complete and reconciled, owner sign-off recorded, source set read-only, launch note sent, hypercare staffed. Departments launch in the wave order set in Phase 5 (pilot first, then the rest), never all on one Monday. Keep the old sources read-only — not deleted — for the whole launch period: read-only is both the rollback and the safety net while people build new habits. The intranet home and hub go fully live with the first department's launch, so every later wave arrives at a structure that already looks inhabited.
Deliverables: per-department launch record (delta reconciled, sign-off, read-only date, launch note sent); hypercare rota for each launch week.
Exit criteria (per department): all users working from SharePoint; source read-only; week-1 friction log reviewed and triaged with no access problem unfixed for more than a day.
Phase 9 — Post-launch governance and optimisation (starts 2 weeks after each launch; monthly rhythm thereafter)
Rollouts decay predictably: sites multiply, permissions drift wider, libraries creep toward the view threshold, and the file share quietly comes back to life as a shadow system. A monthly one-page health check prevents all four: site count vs the site map (any site not on the map gets an owner or a removal date), permission exceptions granted since last review, the largest libraries' item counts against the 5,000-item view discipline, external sharing grants, and the top three friction items from champions. New site requests go back through the Phase 3 process — the tenant that takes any request in month two is the sprawl of Pitfall 2 by month six. Schedule the old sources' deletion decision for 3–6 months after the last launch, once a full business cycle has passed with them read-only.
Deliverables: the monthly health check (sites vs map, permission exceptions, largest-library counts, external shares, top-3 friction); a quarterly access review; the dated decision on deleting read-only sources.
Exit criteria (ongoing): health check produced on schedule; no unowned site older than 30 days; read-only sources deleted only by the recorded decision, after a full business cycle.
Week by week: a 10-week mid-size rollout
The worked schedule below is the mid-size row of the duration table made concrete: a ~150-person organisation, six departments, roughly 1.4 TB of content on file shares plus an old on-premises SharePoint used by two departments, launching department by department.
| Week | Focus | Key outputs | Who is involved |
|---|---|---|---|
| Week 1 | Discovery (Phase 1) | Source inventory started: shares, old SharePoint, personal drives catalogued with owners | Project lead, IT, department heads |
| Week 2 | Content audit (Phase 1) | Numbers per source: 1.4 TB, ~610,000 files, 2.1% over the 400-character path limit, 38% unmodified in 24 months | Project lead, content owners |
| Week 3 | Architecture + governance (Phases 2–3) | Site map signed off (1 hub, 6 department team sites, 4 communication sites); site creation restricted; sharing policy set | Department heads, IT, information risk |
| Week 4 | Build (Phase 4) + cleanup starts (Phase 5) | Hub and all sites built; find-it test passed; departments begin keep/archive/delete in their shares | Project lead, content owners |
| Week 5 | Cleanup + pilot wave (Phases 5–6) | Cleanup cuts migrate set from 1.4 TB to ~820 GB; pilot department migrated with SPMT bulk + delta | Pilot department, project lead |
| Week 6 | Pilot week + training prep (Phases 6–7) | Pilot works solely in SharePoint; friction fixes folded into build checklist; quick references drafted | Pilot department, champions |
| Week 7 | Waves 2–3 (Phases 5, 7, 8) | Two departments cleaned, migrated, trained and launched; sources read-only | Departments 2–3, IT |
| Week 8 | Waves 4–5 | Two more departments launched; old on-premises SharePoint content migrated with permission mapping | Departments 4–5, project lead |
| Week 9 | Wave 6 + catch-up | Final department launched; stragglers and exception content resolved; hypercare across all departments | Department 6, champions |
| Week 10 | Consolidation (Phase 9 begins) | All sources read-only; health-check rhythm set; deletion decision scheduled for month 6 | Project lead, IT, department heads |
Two features of this schedule generalise. The build (Week 4) is one week of ten — discovery, cleanup, pilot and training are the rest, which is why "SharePoint is already licensed, how long can it take?" quotes mislead. And the pilot sits before the department waves, deliberately: every fix it surfaces is applied once to the template instead of six times to live departments.
Worked example: cleanup and the path limit decide the migration plan
Take the Week 1–10 rollout above and do the arithmetic the schedule actually turns on. The Phase 1 inventory counts ~610,000 files / 1.4 TB across the shares. The staleness measure says 38% has not been modified in 24 months — after owners apply the keep / archive / delete rule, the live migrate set falls to ~820 GB: roughly 580 GB that never has to be migrated, permissioned, or trained on, which is the single largest schedule saving in the whole project and it happens with a delete key, not a migration tool.
The path scan is the second calculation. 2.1% of 610,000 files ≈ 12,800 files sit over the 400-character URL limit, concentrated in two departments' deep project folders. At a realistic fix pace — an owner restructuring and renaming their own folders, verified by a re-scan — that is days of work per department in Week 4–5, in parallel with everything else. Discovered in Week 7 during a failed wave, the same 12,800 files would be a launch-blocking incident with the department already told their date. Neither number required a bigger tool or a longer build; both required being measured in Phase 1 instead of discovered in production.
Decision points
Team site or communication site?
- Team site → a defined group creates and edits content together (department working documents, project files). Choose the group-connected kind when the group also needs the shared mailbox, Planner and Teams surface; a team site without a group fits a stable group that collaborates only on files.
- Communication site → a few authors publish finished content to a wide audience (intranet news, policies, forms, HR pages). No group, no shared mailbox, read access for most of the organisation.
- The tiebreaker question: "Will most visitors ever upload or edit a file here?" Yes → team site. No → communication site.
Hub structure
- One hub per major audience or division (typically a single "Intranet" hub for a small/medium organisation), department sites associated to it for shared navigation and rolled-up search.
- Keep hubs flat: a site associates to one hub; resist nesting hubs to mirror the org chart — navigation depth beyond two clicks is where findability goes to die, and the org chart changes more often than the content does.
Migration tooling: SPMT or third-party?
- SharePoint Migration Tool (free, Microsoft) when: sources are file shares and/or on-premises SharePoint; the permission model is being rebuilt anyway (the common case); bulk + incremental passes and CSV/JSON job files cover the scheduling you need.
- A paid third-party tool (e.g. ShareGate; AvePoint or Quest Content Matrix for larger estates) when: permissions and metadata must be carried across with high fidelity, version history must be preserved in bulk, the estate spans many sources needing central reporting, or pre-migration analysis/remediation (path fixes, permission mapping) must happen inside the tool.
- The middle path most mid-size rollouts take: SPMT for the file shares (permissions are being re-expressed anyway), third-party for the old on-premises SharePoint, where version history and existing metadata are the actual content.
Consultant or DIY?
- DIY when: under ~200 staff, sources are file shares plus at most one old SharePoint, someone internal can own the information architecture part-time, and no regulated retention rules apply.
- Bring in a consultant when: the estate passes ~3 TB or several hundred thousand files with permission fidelity required, regulated retention/records rules apply, custom on-premises solutions must be replaced rather than migrated, or no one internally can own the architecture — an unowned architecture is the most expensive outcome of all.
- Highest-value compromise: a few days of expert help on Phases 2–3 only (architecture and governance), then build and migrate internally. A wrong architecture is the costliest thing to fix later; the migration runs themselves are the easy part to staff.
Where SharePoint is the wrong tool
Being clear about fit in Phase 1 prevents the most expensive delay of all: discovering it in the pilot.
- Real-time co-editing of very large spreadsheets and specialist files. SharePoint stores them fine; the working experience belongs to the native application or a specialist system. If a team's core artefact is a 200 MB workbook a dozen people edit simultaneously, fix that workflow separately rather than blaming the rollout.
- A database for structured line-of-business data. Libraries with metadata look database-like, but lookups, transactions and reporting across hundreds of thousands of records belong in Dataverse, SQL or the line-of-business app — with SharePoint at most holding the documents around it.
- External-facing public websites. SharePoint Online is an intranet and collaboration platform for your tenant's users and invited guests; it is not a public web CMS and cannot publish anonymous public sites the way the old SharePoint Server publishing feature once did.
- The only copy of regulated records, unmanaged. If content must satisfy audit-grade retention, disposition and access-evidence requirements, SharePoint can hold it only with the retention and records configuration designed in Phase 3 — the platform alone, defaulted, is not the compliance answer.
What delays this timeline
- Content cleanup of the old file shares. This is the delay in the majority of rollouts: years of duplicates, personal folders on departmental shares, and files nobody will claim. Budget the cleanup effort roughly equal to the migration effort itself, assign it to content owners by name, and start it in Week 1 — a cleanup that starts in migration week finishes after launch week.
- Path-length remediation. Deep folder chains plus long file names against the 400-character limit. Found by scan in Phase 1 it is routine renaming; found by a failed migration job it is a blocked wave.
- Ownership and sign-off latency. Department heads who must approve the site map, the classification rule and their wave's sign-off are the scarcest resource in the plan. Book the Phase 2 and per-wave sign-off meetings during Phase 1, while diaries are open.
- Permission archaeology. File shares whose access was managed by forgotten groups and one-off grants take longer to re-express than the files take to copy. If nobody can say why a folder is restricted, it gets an owner interview, not a guess.
- Scope added mid-rollout. Every "while you're at it, rebuild the intranet home page / the forms library / the board papers process" restarts Phase 2 for that piece and invalidates the pilot's findings for the changed parts. The later-list from Phase 1 is the pressure valve — use it.
- Launching into peak season. A flawless structure launched into a department's busiest fortnight gets blamed for the fortnight. Move the wave, not the training.
Pitfalls that actually break SharePoint rollouts
- Migrating the mess. Lifting the file share exactly as it stands — duplicates, dead content, nine-level folders — produces a SharePoint that is the file share with worse navigation, and users conclude the platform is the problem. Cleanup is not preparation for the rollout; it is half the rollout.
- Site sprawl from day one. Leaving site creation open to everyone "to encourage adoption" yields dozens of unowned, overlapping sites within a quarter, after which findability and governance are both gone. Restrict creation in Phase 3 and make the request process fast — controlled is not the same as slow.
- Copying the folder structure instead of designing the architecture. Folders hide content from search, metadata and views; a migrated folder tree wastes the platform. Phase 2's site/library/metadata design is the fix, and the 5,000-item view threshold will punish skipping it.
- Unique permissions everywhere. Breaking permission inheritance folder by folder recreates NTFS chaos inside SharePoint and makes access reviews impossible. Secure at site level; treat every unique permission as a documented exception (see the permission model in Phase 3).
- Big-bang launch. Migrating everyone on one weekend concentrates every permission surprise, training gap and missing file into one Monday. Waves exist so the second department benefits from the first department's friction log.
- The shadow file share. Leaving the old share writable "just in case" guarantees two versions of everything within a month. Read-only at cutover is non-negotiable; deletion later, by decision, is fine.
- No owner after launch. Sites and the hub rot at a predictable rate once the project team disperses. Phase 9 works only with named owners per site — "IT" is not an owner for departmental content.
References
Primary vendor documentation consulted for the limits and platform facts in this timeline:
- Microsoft Learn — SharePoint service limits: URL path length (400 characters), list view threshold (5,000 items), items per list/library, maximum file size, storage per site and tenant storage allowances: https://learn.microsoft.com/en-us/office365/servicedescriptions/sharepoint-online-service-description/sharepoint-online-limits
- Microsoft Learn — SharePoint Migration Tool (SPMT) documentation: supported sources (file shares and on-premises SharePoint), bulk migration with CSV/JSON job files, and incremental migration behaviour: https://learn.microsoft.com/en-us/sharepointmigration/migrate-to-sharepoint-online
Read next
- Moving from SharePoint to Google Workspace Checklist — if the destination changes mid-evaluation, the enterprise checklist for moving off SharePoint, with named tools and the limits that apply on that route
- SharePoint rollout plan for older employees — the audience-specific companion: permission-group naming and admin-centre steps for a less technical workforce
- Airtable implementation timeline — the same week-by-week planning pattern applied to an Airtable rollout, with its plan limits driving the schedule
Frequently asked questions
How long does a SharePoint implementation take?
For a small organisation standing SharePoint up alongside a modest file share, 4–6 weeks. A mid-size organisation (50–200 staff) migrating several departments runs 8–12 weeks. Larger, permission-heavy or multi-unit estates run 12–20 weeks or more. In every case the build itself is the minority of the calendar — the content audit, cleanup, pilot and training set the pace.
What takes the longest in a SharePoint rollout?
Content cleanup of the old file shares, by a wide margin. In the worked example above, owners applying a keep / archive / delete rule cut the migration set from 1.4 TB to about 820 GB — work done by department staff reviewing their own content, which no tool can do for them. Second is permission archaeology: re-expressing years of file-share access decisions as a deliberate site-level permission model.
Do we need a third-party migration tool, or is Microsoft's free one enough?
For file shares — the common case — the free SharePoint Migration Tool is usually enough: it supports bulk and incremental runs, and file-share permissions are normally being rebuilt to the new model anyway. Budget for a third-party tool (ShareGate is the common choice at this scale) when an old on-premises SharePoint's permissions, metadata and version history must come across with fidelity, or when you need central reporting across many sources.
What are the SharePoint limits most likely to bite during implementation?
Two. The 400-character URL path limit, which long file-share folder chains routinely exceed (about 2% of files in the worked example, ≈12,800 files to fix), and the 5,000-item list view threshold, which breaks unfiltered views on large libraries unless columns are indexed and views are designed for it in the build phase.
When does the old file share get switched off?
Read-only at each department's cutover — that part is immediate and non-negotiable, or the shadow-system problem starts. Deletion comes 3–6 months after the final launch, by a recorded decision, once a full business cycle has passed with the shares read-only and nothing has had to be recovered from them as the working copy.