Key Takeaways
- Migrating organizational knowledge isn't just about moving files—it's about curation and findability.
- Ruthlessly purging outdated content ("cruft") prevents you from migrating problems into a new system.
- Establishing clear ownership for every piece of content ensures nothing goes stale after migration.
- Organizing for search and retrieval—not folder structure—makes content actually usable.
- A well-organized knowledge base is inherently AI-ready, letting employees ask questions and get real answers.
You know the state of your documents is a problem. Everyone knows.
Files in SharePoint. Files in Google Drive. Files on the old shared drive that nobody's touched in three years, but nobody will delete. Wikis with outdated content. Notion pages that one team uses. Confluence spaces from before the reorg. PDFs in email attachments that never made it anywhere permanent.
The knowledge exists. It's just scattered across a dozen systems, in a hundred formats, maintained by nobody in particular. And every time someone new joins the company, they experience the chaos firsthand: asking around, searching fruitlessly, eventually interrupting a colleague who "just knows where things are."
At some point, someone says: We need to consolidate everything into one place.
They're right. But "move everything into the new system" is where most migrations go wrong. Because moving a mess from five places into one place just gives you a mess in one place.
Here's how to do it properly.
Step 1: The Honest Inventory
Before you migrate anything, you need to know what you actually have. This sounds obvious, but most organizations skip this step because it's tedious. Don't skip it. For a detailed framework, see our guide on how to audit company knowledge.
Create a simple spreadsheet. For each knowledge source (SharePoint, Drive, wiki, etc.), document:
- What's there: Types of content (policies, procedures, how-tos, templates, meeting notes, project docs)
- How much: Rough volume (hundreds of files? thousands?)
- How old: When was it last meaningfully updated?
- Who owns it: Is there a person or team responsible?
- Who uses it: Does anyone actually reference this content?
This inventory will be sobering. You'll discover that 60-80% of your content is outdated, redundant, or orphaned. That's normal. That's the point of the inventory—to see the problem clearly before trying to solve it.
Step 2: The Ruthless Purge
Warning: This is where most migrations stall. People are reluctant to delete anything because "someone might need it." That instinct, while understandable, is exactly what created the mess in the first place.
Set clear criteria for what gets purged:
- Not updated in 2+ years and not a foundational policy? Archive or delete.
- Duplicate versions with no clear single source of truth? Keep one, delete the rest.
- Meeting notes and project docs from completed projects? Archive.
- Content nobody claims ownership of? Flag it. If nobody steps up in 30 days, archive it.
Archiving doesn't mean migrating. It means putting content in cold storage where it won't clutter your new system but can be retrieved if truly needed. Most archived content will never be touched again, and that's fine.
Step 3: Prioritize High-Value Content
Not all surviving content is equally important. Prioritize what gets migrated first based on impact:
Tier 1 — Migrate immediately: Onboarding materials, core processes, product documentation, compliance and policy documents, customer-facing knowledge.
Tier 2 — Migrate soon: Department-specific procedures, templates, job aids and reference guides, training materials.
Tier 3 — Migrate if time permits: Historical project documentation, archived communications, reference materials that are "nice to have."
Tip: Start with Tier 1. Get it right. Let people use it and give feedback. Then move to Tier 2. This iterative approach prevents the "big bang" migration that overwhelms everyone and typically fails.
Step 4: Organize for Retrieval, Not Storage
This is the most important step and the one most organizations get wrong. Traditional file organization mirrors how people create content: by department, by project, by date. But that's not how people find content.
Nobody thinks: "I need the Q3 2024 version of the procedure that lives in the Operations folder under the Process Documentation subfolder." They think: "How do I process a customer refund?"
Organize your migrated content around:
- Topics and tasks, not departments
- Questions people ask, not document titles
- Workflows and processes, not org chart hierarchies
The AI-Ready Structure
If you're planning to layer AI on top of your knowledge base—and you should be—structure matters even more. AI retrieval works best when:
- Each document covers one clear topic
- Content has descriptive titles and headers, not "Document_v3_final_FINAL"
- Information is written in plain language, not jargon-heavy shorthand
- Metadata (tags, categories, descriptions) is consistent and accurate
Think of it this way: if a new employee couldn't find and understand a document within 60 seconds, it's not organized well enough—for humans or AI.
Step 5: Assign Ownership (Or Don't Migrate)
Warning: Content without an owner is content that will go stale. If you migrate it without assigning ownership, you're just recreating the same problem in a new system.
Every piece of migrated content needs:
- An owner: A specific person (not a team, not a department—a person) responsible for keeping it current
- A review cadence: Quarterly for fast-changing content, annually for stable policies
- A sunset criteria: Under what conditions should this content be archived or retired?
If you can't identify an owner for a piece of content, don't migrate it. Put it in the archive. If someone needs it later, they'll surface it—and that person becomes the owner.
Step 6: Choose Your Migration Method
Not all content migrates the same way. Here are four approaches, roughly ordered from most to least effort:
- Rewrite and restructure: Best for high-value, high-use content that's poorly written or organized. Takes the most time but produces the best results.
- Edit and import: Content is fundamentally sound but needs cleanup—updated formatting, removed jargon, added metadata. Good for Tier 1 and Tier 2 content.
- Bulk import with tagging: Content is already well-written and just needs to be moved and tagged in the new system. Rare, but it happens.
- Link and reference: Some content (like regulatory documents or vendor specs) doesn't need to live in your knowledge base. Just link to the authoritative source.
Most organizations will use a mix of all four methods. The mistake is treating everything as a bulk import when most content needs at least some editing.
Step 7: Test Findability
Before you declare the migration complete, test whether people can actually find what they need. Grab five employees who weren't involved in the migration. Give them ten real questions they'd need to answer in their jobs. Watch them search.
If they can't find the answer within a minute or two, your organization isn't done. It doesn't matter how neatly organized the back end is—if the front-end experience is "I still can't find anything," the migration has failed.
Tip: Test with new employees or cross-departmental colleagues who lack context. If they can find answers, anyone can. If they can't, your content needs better titles, tags, or structure.
Step 8: Manage the Transition
The technical migration is half the battle. The human migration is the other half. People have habits. They'll keep going to the old systems unless you actively manage the transition.
- Set a hard cutoff date for the old system. Communicate it early and often.
- Redirect, don't just announce. When someone asks "Where's the refund policy?" don't just say "It's in the new system." Send them the direct link.
- Make the new system the path of least resistance. If it's easier to find things in the new system, people will use it. If it's not, they won't—no matter how many emails you send.
- Celebrate small wins. When someone finds something quickly in the new system that would have taken 20 minutes before, share that story.
The "Done" Myth
Knowledge migration is never truly done. Content changes. People leave. New processes emerge. The difference between a living knowledge base and a digital graveyard is ongoing maintenance—the ownership model and review cadences you established in Step 5.
Build the habit of continuous curation into your team's workflow. A little maintenance every week prevents another massive migration project in three years. Using workflow assistants can help automate parts of this ongoing maintenance.
One More Thing: AI
If you've done the work described above—audited, purged, organized, tagged, and assigned ownership—you've also done 90% of the work needed to make your knowledge base AI-ready. A well-structured knowledge base is exactly what AI needs to provide accurate, contextual answers to employee questions.
Instead of searching through folders or asking colleagues, employees can simply ask a question and get an answer drawn from your organization's actual, curated knowledge. This is the promise of the AI knowledge assistant.
JoySuite can be the destination for your migration—or the AI layer on top of wherever your content ends up. Employees ask questions, Joy finds answers across your documents. The work you put into organizing your knowledge becomes instantly accessible.
