Knowledge management without a knowledge management team.

Knowledge management at small firms tends to look like this: a folder called Precedentsthat someone organised in 2019, twelve subfolders deep, half empty, nobody's sure if anything in it is current. The result is fee-earners drafting from scratch what colleagues finished last quarter, and reviewing case law that someone else researched three months ago.

The fix isn't a knowledge management system. It's three small disciplines run by named people: a precedent bank that's actually maintained, a decision log that captures the reasoning behind unusual calls, and a culture of searching before drafting.

The precedent bank that gets used

The folder of precedents nobody opens has one or more of three problems: it's out of date, it's unsearchable, or nobody's sure what's authoritative. Fix each:

Out of date — assign a per-precedent owner

Every precedent has a named owner — a fee-earner who is responsible for keeping it current. Annual review on a rolling schedule. If a precedent has no owner, it gets archived; the absence of an owner is itself evidence that the precedent isn't valuable enough to maintain.

Unsearchable — flat structure plus a tagging discipline

Twelve folders deep is unsearchable. The version that works: a flat folder per matter type, with files named in a consistent pattern (NDA — mutual — 2025-02 — v3.docx). Plus a one-line description on the file properties. Most modern document management systems search the description; users can find a precedent in 15 seconds instead of giving up.

Authoritative? — mark each precedent

Three states, marked clearly in the filename or first line of the document: master (current, approved), working (recent draft, may need review), reference (old, kept for context only). Without this distinction, fee-earners default to the most recent file regardless of its status.

The decision log

The single highest-leverage knowledge tool at most small firms — and the one nobody builds. The format is simple: when a fee-earner makes a non-obvious call on a matter, they write two paragraphs in a shared log:

• The situation (anonymised if relevant) and the question that came up.
• The decision, the reasoning, and the alternatives that were rejected and why.

That's it. Three minutes per entry. Twelve months in, you have a hundred entries that future fee-earners re-read when the same situation comes up — and instead of rebuilding the analysis from scratch, they read the existing reasoning, agree or refine, and move on.

The discipline that makes this work: it's on the weekly matter review agenda (see matter tracking sheet for the wider review structure). Every fee-earner adds at least one entry from the past week, or explicitly says they had no decisions worth logging. The accountability is the lightest possible — it just stops the log from atrophying.

The search-before-you-draft rule

A cultural rule, not a technical one. Before drafting any substantial document, the fee-earner spends 90 seconds searching the precedent bank and the decision log for the nearest match. Three outcomes:

• Match found — start from it, edit. Saves 30–60 minutes on most matters.
• Partial match found — start from it, rework. Saves some time, plus surfaces the gap in the precedent that the fee-earner can fix as part of the work.
• No match found — draft from scratch, then add the result to the precedent bank. Net cost today; net benefit for everyone next time.

The rule sounds obvious, but it's under-practised at most firms because the search infrastructure is bad enough that fee-earners have learned not to bother. Fix the search, and the habit follows.

What not to build

Three things small firms over-invest in around knowledge management:

• An internal wiki nobody updates. The decision log does most of what a wiki tries to. Wikis drift; logs accrete chronologically and stay honest.
• A formal taxonomy.Tagging schemes that require fee-earners to choose from 30 categories don't get used. Three or four matter-type buckets are enough.
• An expensive knowledge management product. For firms under 50 fee-earners, the cost of dedicated KM software rarely pays back. The discipline is what pays back; the software is a multiplier on the discipline. Build the discipline first.

The induction integration

New joiners in the first two weeks: half a day spent reading the decision log from the past 12 months. Less to memorise; more to absorb how the firm thinks about non-obvious calls in their practice area. Most firms induct on systems and forget to induct on judgement — which is the harder thing to transmit.

The training pathway article (training pathway from paralegal to lawyer ) covers the wider development picture. The decision log is the single highest-leverage tool inside it.

The annual cleanup

Once a year, half a day per practice area lead:

1. Review the precedent bank for the practice area — archive what's out of date, identify gaps, assign new owners where needed.
2. Skim the decision log entries from the past year — flag three or four that are sufficiently common they should be folded into a precedent or a checklist.
3. One paragraph note to the team on what changed and what to look at.

Three or four hours, once a year, per practice area. Lower cost than a single matter going badly because someone couldn't find the precedent.

What good looks like at year one

Every precedent has an owner and a status. The decision log has 80–150 entries that get genuinely re-read by fee-earners working on similar matters. New joiners come up to speed faster because they have the firm's accumulated reasoning in front of them, not just its templates. Fee-earners spend more of their drafting time on the bits that actually need their judgement, less on recreating standard structure that already exists in the bank.

Underrated outcome: the firm's knowledge stops being locked in the heads of three senior fee-earners. When one of them leaves, what they knew leaves slower than it otherwise would.