A knowledge base is where your staff go to answer “How do I…?” How to process a refund, request a laptop, onboard a new team member, or handle the situation that only comes up twice a year. When it works, people check it first. When it doesn’t, they ask a colleague, and the colleague becomes the knowledge base.
Most organizations already have one, or several. The trouble usually isn’t the tool. It’s that:
- Nobody’s sure which version of a procedure is the current one.
- Articles were written once, years ago, by people who have since left.
- Search returns 400 results, and the right one is on page three.
- Every team writes differently, so each article has to be decoded before it’s useful.
- Nobody is responsible for keeping it current, so nobody does.
- You’d like an AI assistant to answer staff questions, and you know it would be answering from all of the above.
We help organizations build knowledge bases, or rebuild the ones they have, so staff can trust what they find.
How we approach it
A knowledge base is only as good as its upkeep. We spend as much time on how content gets written, reviewed, and retired as on how it’s organized.
- Start with the questions people ask. We look at what staff search for, what they ask colleagues, and what the help desk hears every week. That shapes what the knowledge base needs to cover, and what it doesn’t.
- A structure that matches the work. Categories, templates, and metadata organized around tasks and roles, so people can browse as well as search.
- Clear, consistent writing. Article templates and a short style guide, so a procedure written by finance reads like one written by HR. Where it helps, we rewrite or edit the content ourselves.
- Cleanup before migration. What to keep, update, merge, or delete before anything moves. Usually a lot less moves than people expect.
- Owners and review dates. Every article has someone responsible for it and a date when it gets checked. Simple routines your team can sustain.
- Ready for AI. Clear, current, well-structured articles are exactly what AI assistants need to give good answers, and what keeps them from confidently giving bad ones.
- Measuring what’s used. Search terms, page views, and failed searches show where the gaps are. We help set that up so the knowledge base keeps improving after we leave.
Knowledge base or intranet?
They overlap, and many organizations keep both in the same platform. The difference is the job:
- A knowledge base answers questions. Procedures, policies, how-to guides, and reference material. People come to it with a specific need.
- An intranet connects people. News, announcements, team pages, and a front door to everything else. People come to it to stay informed and find their way around.
If your challenge is more about the front door, see our SharePoint and intranet consulting. If you’re not sure where the problem is, a knowledge management assessment can help sort that out.
The tools
We work in what you have. Most of our knowledge base work is in SharePoint and Microsoft 365, and we’ve also worked in Confluence, Notion, Google Sites, Airtable, and custom wikis. If your current tool really isn’t a good fit, we’ll help you choose what comes next. We don’t sell software, and we don’t have a favorite to push.
What this has looked like
A global operations knowledge base. A Fortune 500 technology company’s operations team was working from an outdated, on-premises knowledge base with more than 1,000 documents. We interviewed nearly 50 staff, designed a new structure and taxonomy in SharePoint Online, built templates, and rewrote the articles for clarity and consistency. We also set up built-in translation for staff around the world. The new knowledge base went live within six months, and within a year the client’s own team was running it, with subject matter experts updating their own content. Read the full story →
A habit of writing things down. For some clients, the most useful thing we’ve built isn’t content but a program: SOP templates, owners, review routines, and the change management that gets staff writing and updating procedures themselves. More on that work →
Questions people usually ask first
Do we need a new platform?
Usually not. Most knowledge base problems are about content and ownership, not software. We start with what you have and recommend something new only if it clearly earns its place.
Will you write the content, or just set things up?
Either, or both. Some clients want a structure, templates, and a style guide their staff use to write. Others want us to rewrite existing content, as we did for more than 1,000 articles on the project above.
How do we keep it from going stale again?
Owners, review dates, and a simple way to flag problems. We also look at usage data, so you can see which articles matter most and focus your upkeep there.
Can we use an AI tool instead of a knowledge base?
AI tools don’t replace a knowledge base; they rely on one. An AI assistant is only as accurate as the content it draws on. A well-maintained knowledge base is the best preparation for AI. See how we help with AI rollouts.
For tips on measuring what’s working, see our post on knowledge base analytics, or read about our broader knowledge management work.
Tell us about your knowledge base
A few sentences is plenty: what you have now, what’s frustrating about it, and what you’d like it to do. We read every message ourselves and reply within one to two business days.
Prefer email? Write to us at [email protected].
