Knowledge Management Needs a Makeover
Someone asked us a good question last week:
Is "knowledge management" still the right label?
KM doesn't get a lot of love at many law firms. The usual complaint is that it sounds boring, and worse, detached from revenue — KM is the thing lawyers move into when they are done with practising law. Your life is a ceaseless declining spiral of writing precedents, conducting research, and filing documents.
Obviously, we don't think that is right. We think KM has become more important in the last few years, but the problem is that the legal world has moved beyond conventional "knowledge management" and the label hasn't quite followed the evolution of the job.
Say "data strategy" and watch what happens
If we take the exact same tasks that a "knowledge management" team does — the same discipline, the same people, the same work — and call it "data strategy", suddenly the view shifts. Everyone nods:
Of course data is important. We love data. Shut up and take my money!
It is as if KM never hired an ad agency, and the data folks went out and got the team from Sterling Cooper.
None of that is rational and all of it is real.
Turn that pyramid upside-down
Most of us who have worked in legal carry some version of the value pyramid around in our heads:
data becomes information,
information becomes knowledge, and
knowledge becomes insights the moment a lawyer applies it to an actual matter.
Knowledge sits high up on that pyramid. Being high in the pyramid means it is narrow, it is distilled, and it is removed from the source. Curated precedent banks, practice notes, beautifully maintained know-how sites, all built for one human to read one document at a time.
When legal work was performed by humans, KM supported the lawyer who was the only thing capable of doing the last step. Except now, the machine has inverted the pyramid. KM is standing at the wrong place when machines can do a lot of the work. Search and curation is now "free" with AI. Leverage and value is moving down the pyramid towards the raw materials, because the raw materials can be processed, whenever needed, and delivered "just-in-time". The old model was “just-in-case” — curate it now, in case someone needs it later. That disappears when AI can do the work for free.
The market has already started changing the language of KM to match this new paradigm:
we have already stopped saying "precedents" and started saying "examples",
we have stopped saying "templates" and started saying "standards".
The old established functions still say "precedents" and "templates", and the teams that formed during the AI wave say "examples" and "standards". They may seem like small words, but they encode a different theory of what documents do.
Just a new name is not enough
It won't be enough to just rebrand knowledge management into data strategy. More work has to be done to follow where the value is shifting. The new KM role has to leverage the new capabilities of the tools — and it has to hold onto the relationships and structure inside the documents, because feeding a machine a pile of disconnected pieces is not a data strategy (at least, not a good one).
Our hunch on how it plays out in the next few years is that AI will go wide — scattergunned onto every use case, everything at once — and then it will narrow onto the 20% of use cases where it actually works and delivers value. Getting anywhere near 20% means someone (or something) has to go and scrape the right pieces of data among the multitude of systems, so the machine can be fed the right context when needed. When this happens, the "data" team becomes a permanent department at firms and legal departments. At that point, knowledge management stops being a curation and refinement role, and becomes an architecture and system role.
What should KM do today?
If you work in KM, we believe the immediate steps today are:
Do not solve this by buying something. We keep meeting teams running five overlapping AI platforms who are evaluating a sixth to bring the other five together. If you are drowning, another cup of water is probably not the answer.
If your knowledge database is built for a human to read one document at a time, nobody will recognise what it is worth until they are stranded and in trouble. Your data needs to be made ready for machines.
If your data set is already structured so a machine can navigate it — relationships and provenance intact — then stop calling it knowledge. Call it data. In-house legal teams especially: legal has never been funded for a knowledge specialist, but the organisation already has a data team with standards and headcount that covers everything except legal. Take it to them, and let the new label do some work.
All the hard work that KM teams have been doing for decades — taxonomy, curation, provenance, knowing which document is the good one and why — is the thing that makes data valuable to AI. The broader legal market is still catching up, and the risk is that many lawyers will not come to the same view unless the KM folks do the work on the packaging, and change the label. The job of KM teams has never mattered more than it does right now. It just needs to start sounding like it.
