Every law firm I've ever spoken to says knowledge management is important. And almost every law firm I've ever worked with has a KM system that nobody uses properly.
That's not a contradiction - it's the whole problem. KM has been on the agenda at mid-market law firms for the better part of twenty years. Partners nod along in strategy days. Someone gets tasked with "looking into it." A platform gets bought, maybe even implemented. And then, slowly, it withers. Templates go stale. Precedents don't get tagged. The know-how database becomes a graveyard of documents from 2019 that nobody trusts enough to rely on.
We've tried KM before. Lawyers won't contribute to knowledge bases - they're too busy billing.
Fair. And you're not wrong about the symptom. Lawyers are busy billing. Their time is carved into six-minute increments, and "updating the precedent library" doesn't have a client code attached to it. But something has changed that makes this more than an institutional tidiness problem: knowledge management is now the foundation for everything AI promises to do for your firm. If that foundation isn't there, you'll spend a lot of money on AI tools that underperform - or worse, produce outputs nobody trusts.
That shift - from KM as a "nice-to-have" to KM as a commercial imperative - is what this piece is about.
If the only reason to fix your knowledge management was "it would be nice if associates could find things faster," I probably wouldn't be writing this. Firms have survived with messy KM for years. Partners kept the knowledge in their heads, associates learned by osmosis, and the whole thing sort of worked - expensively and inefficiently, but it worked.
What's different now is AI.
The firms getting genuine value from AI in legal services - document review, precedent search, automated first-draft generation, client service triage - are overwhelmingly the ones that had their knowledge organised before they plugged AI into it. I've written separately about how law firms are using AI for knowledge management, and the pattern is consistent: the technology isn't the bottleneck. The data is.
Think of it this way. If you ask an AI tool to find the most relevant precedent for a specific commercial lease dispute, it can only search what it can see. If your precedent documents are scattered across individual partner drives, saved in inconsistent formats, tagged with whatever metadata someone felt like adding at the time - or not adding - the AI isn't going to magically produce useful results. Garbage in, garbage out. Except now the garbage costs you a six-figure AI licence fee on top.
We've seen this play out. A 300-person firm we worked with last year had invested in what was, on paper, a perfectly decent AI-powered research tool. I was in the room when the head of IT walked us through the usage data - six months in, adoption had flatlined. You could see it in the graph: a spike in the first few weeks, then a long, slow decline to almost nothing. When we dug into why, the answer was depressingly predictable. The firm's document management was so inconsistent that the AI kept surfacing irrelevant results. Associates tried it, got burned a couple of times, and went back to asking the partner down the hall. Sixty grand a year on a tool that was functionally a very expensive search bar.
I've written a companion piece on three things a 300-person law firm can do with AI this quarter - and it's no coincidence that KM improvements feature as the first practical step.
"Knowledge management" means different things to different people, and that vagueness is part of why it's been so easy to deprioritise. So let me be specific.
For a mid-market firm - say, 80 to 400 fee earners - practical KM is less about grand systems and more about a handful of disciplines that actually hold together.
The starting point is document management that people actually use. Not just a DMS that exists, but one where documents are saved consistently, named intelligibly, and retrievable without needing to know the exact matter number or the name of the partner who worked on it. This sounds basic. It is basic. And yet I've sat in firms where finding a specific contract from two years ago requires three phone calls and a favour. That's not a technology problem. That's a habits problem.
From there, precedent tagging matters more than most firms realise. Your firm has done thousands of transactions. Somewhere in those files are precedents for almost anything a client could ask for. The question is whether anyone can find them in less time than it takes to draft from scratch. Tagging by matter type, jurisdiction, complexity, and outcome - not just client name and date - is what makes the difference between a precedent library and a very large pile of documents.
Template libraries are only useful if they reflect current law and current firm standards. A template that hasn't been reviewed in eighteen months is a risk management problem, not a productivity tool. Someone needs to own the update cycle. That's a role, not a task.
There are two other things that tend to get overlooked. The first is expertise location - in a firm of 200 people, it's genuinely difficult to know who's done what. A partner in the Manchester office might have handled exactly the type of cross-border IP dispute that's just landed in London, but unless there's a way to surface that experience, the London team starts from scratch. Simple internal profiles and updated matter records make a surprising difference. The second is know-how: legal updates, case notes, internal briefings, training materials. Most firms produce some of this. Very few make it easy to find.
None of this is revolutionary. That's sort of the point. The firms that do KM well aren't using exotic technology - they're doing the boring bits consistently.
Right, here's where it gets uncomfortable.
The reason most KM initiatives fail in law firms isn't technology. It's behaviour. Lawyers are trained to be individually excellent. Their incentive structures reward personal billing, personal client relationships, personal expertise. Sharing knowledge in a structured, systematic way runs slightly against the grain of how most lawyers have been rewarded throughout their careers. Why spend twenty minutes properly tagging a precedent when those twenty minutes could be billed?
That's not laziness. It's rational behaviour within a system that doesn't value KM contribution.
So if you're a managing partner or COO reading this and thinking "we just need to get the lawyers to use the system" - I'd push back gently. You need to design the system so that using it is easier than not using it. That's a fundamentally different design challenge.
What does that look like in practice? A few things.
Make contribution effortless. If saving a document to the KM system requires fifteen clicks, a metadata form with twelve fields, and a taxonomy selection from a dropdown menu that hasn't been updated since 2020 - people won't do it. Build contribution into existing workflows. Auto-tagging. Smart defaults. Integration with the DMS so documents flow into the knowledge base without a separate step.
Make retrieval obviously better. If the KM system returns worse results than asking the associate in the next office, nobody will use it. Invest in search quality. When a lawyer searches for "restrictive covenant dispute, financial services, High Court" and gets back three directly relevant precedents in under ten seconds, they'll come back tomorrow.
Recognise contribution. This doesn't need to be dramatic. Some firms include KM contribution in performance reviews. Others give PSLs authority to flag gaps and request specific contributions from fee earners. One firm I know does a quarterly "most useful precedent" callout in their all-hands - yes, it's a bit naff, but the head of real estate there told me it's the only KM initiative that's ever actually changed behaviour. The point is that if the only thing that gets measured is billable hours, KM will always lose.
Start with the pain points. Don't try to build a comprehensive knowledge base from scratch. Find the specific area where people are wasting the most time reinventing the wheel - employment tribunal bundles, commercial lease templates, regulatory filings - and fix that first. A small, visible win does more for adoption than a grand strategy document.
I remember working with a firm where the real estate team was spending, on average, four hours per transaction recreating standard lease documentation. Four hours. Not because the precedents didn't exist, but because they were buried across individual partner directories with no consistent naming convention. We helped them build a structured template library with a decent search layer. Within three months, that four hours had dropped to about forty-five minutes. The head of real estate - who had been openly sceptical when we started - became the biggest KM evangelist in the firm. Not because anyone told her to. Because her team got their evenings back.
This is where firms often go wrong. They start with the technology decision - "should we buy iManage or NetDocuments?" - before they've worked out what they actually need the technology to do.
The tools that matter for legal KM fall into a few categories, and it's worth being clear about what each one does and doesn't solve.
Document management systems like iManage and NetDocuments dominate the legal market for good reason. Both support matter-centric filing, version control, and integration with the tools lawyers already use - Outlook, Word, practice management systems. The choice between them is less about features and more about your existing technology landscape, your hosting preferences, and honestly, which one your IT team is more comfortable supporting.
External knowledge platforms - Lexis+, Practical Law - provide current law, drafting notes, standard forms. Useful, but they're not a substitute for internal KM. Your firm's specific precedents, your approach to particular transaction types, your templates - that's the stuff that gives you a competitive edge, and it lives inside your systems, not in a third-party database.
Search and retrieval is increasingly where AI earns its keep. Tools that can search across your DMS, emails, and knowledge base using natural language queries - not just keyword matching - are genuinely changing how lawyers find information. Some firms are building this with Microsoft Copilot if they're in the Microsoft ecosystem; others are using specialist legal AI tools. But the search is only as good as the content it's searching. Which brings us back to foundations.
Workflow and automation tools - contract automation platforms like Documate or Hotdocs, practice management systems, matter management platforms - all play a role in KM by structuring how work gets done and how outputs get stored. The more structured the workflow, the more knowledge gets captured as a byproduct rather than requiring a separate contribution step.
The budget question is real. A full KM technology stack for a 200-person firm can run anywhere from £50k to £250k annually, depending on what you already have and how much you need to add. But the biggest expense isn't usually the software - it's implementation and change management. A £30k platform that people actually use beats a £150k platform that sits empty. Every time.
Here's where I want to pull the threads together, because this is the commercial argument that should be in your next board paper.
AI in legal services is moving fast. Document review, due diligence, contract analysis, regulatory monitoring, client communication triage - these are all areas where AI is already delivering measurable results in firms that have the foundations in place. A professional services firm we worked with reduced proposal turnaround time by 60% using AI-powered document generation. That wasn't because they bought a clever tool. It was because they'd spent the previous year getting their knowledge organised - consistent document structure, tagged precedents, current templates - so the AI had something coherent to work with.
Firms that skip the KM step and go straight to AI adoption are the ones who end up in my inbox twelve months later, frustrated that their AI investment isn't delivering. At that point, they're paying for both the AI tool and the retrospective KM work they should have done first. It costs more, takes longer, and generates a healthy dose of internal scepticism about AI that makes the next initiative harder to get approved.
The sequence matters: get your knowledge organised, then layer AI on top.
But we don't have time to sort out KM before we start on AI - the partners are already asking about it.
I hear this a lot. And I get it - there's real pressure to show AI progress. But you can do both in parallel if you're smart about it. Start AI in one contained area where your knowledge is already reasonably organised. Use that as the proof of concept. Meanwhile, fix the KM foundations in the areas where AI will matter most. You're not choosing between them - you're sequencing them.
If you're thinking about where your firm sits on this, we've put together an AI readiness assessment specifically for legal services that covers data foundations, infrastructure, governance, and cultural readiness. It's a useful starting point for an honest conversation about what needs to happen before the AI conversation becomes productive.
I'm wary of articles that diagnose problems without offering anything actionable. So here's what I'd suggest if you're a managing partner, COO, or head of IT at a mid-market law firm thinking "right, we probably need to sort this out."
Audit what you've got. Not a six-month project - a two-week exercise. Where do documents actually live? How are they organised? What's in the DMS versus individual drives versus email attachments? How current are your templates? You'll probably find the answer is worse than you expected, and that's fine. At least you know.
Pick one practice area. Don't try to fix KM across the whole firm at once. Pick the team that's most receptive and has the most obvious pain point. Build something that works for them. Then use that as the proof of concept for wider rollout.
Assign ownership. KM without an owner dies. Whether that's a dedicated KM professional, a PSL with an expanded remit, or a partner champion - someone needs to be accountable for it. Not as a side project. As part of their actual role.
Connect it to the AI conversation. If your firm is having any kind of discussion about AI - and in 2025, almost every firm is - frame KM as the prerequisite, not a separate initiative. The budget conversation is easier when KM is positioned as "AI readiness" rather than "let's tidy up our filing." I've written about how to build the business case for this kind of investment - framing it as AI readiness rather than just knowledge sharing makes a material difference to how the board receives it.
Keep it simple. The firms that do KM well are not the ones with the most sophisticated systems. They're the ones where the basic disciplines - consistent saving, sensible naming, regular template reviews, searchable precedents - happen reliably, every day, across the firm. Boring, consistent, effective. A bit like good legal advice, really.
KM is the foundation for AI in legal services. If you're wondering what else needs to be in place, our AI readiness content for legal firms maps out the full picture, from data infrastructure to governance to cultural readiness.
The firms that get this right over the next twelve to eighteen months will have a genuine advantage - not just in efficiency, but in their ability to adopt AI tools that actually work, serve clients faster, and retain the institutional knowledge that walks out the door every time a senior associate moves firms. The ones that keep putting it off will find, eventually, that the cost of catching up is considerably higher than the cost of starting now.