There’s a question I use as a scalpel:
“If the person who ‘knows’ isn’t here tomorrow, what breaks—exactly?”
When knowledge lives in heads, the business works… until it doesn’t. And that “until” usually shows up at the worst time: a demand peak, a critical incident, an unexpected turnover, a new location, a new supplier, a new tool.
I’m not interested in “documenting for the sake of it.” I’m interested in operational continuity: turning experience into something searchable, transferable, and governed.
The real problem isn’t a lack of documents. It’s dependence on memory.
Many companies “have things written down.” Folders, PDFs, email threads, decks, chats.
That’s not knowledge management. That’s stored noise.
Knowledge management starts when I can answer—consistently—questions like:
-
- what is the official, current way to do this?
-
- why do we do it this way?
-
- what exceptions are allowed?
-
- what decision was made, under what criteria?
-
- where does learning go when something fails?
If I can’t answer, what I have isn’t knowledge. It’s scattered information.
## SOPs aren’t the same as operating playbooks (and neither equals knowledge)
SOPs usually describe the “what” and “how”.
A strong operating playbook also includes:
-
- the why (criteria)
-
- the when not (boundaries)
-
- the what if (exceptions)
-
- and the change trace (what was updated, and why).
Knowledge management goes further: it captures tacit and contextual learning from real execution.
Simple line: SOPs describe actions; knowledge supports decisions.
## What’s at stake: margin, quality, and learning speed
When knowledge isn’t managed, costs appear everywhere:
-
- slow onboarding
-
- rework caused by different interpretations
-
- dependency on “bridge people” between areas
-
- repeated decisions because criteria are forgotten
-
- constant escalation: everything ends up as “ask someone.”
It burns time, yes. But it burns margin and calm even more.
## Early symptoms that knowledge is evaporating
-
- the same questions repeat weekly
-
- answers depend on who replies
-
- key people can’t disconnect without chaos
-
- changes are shared in messages but never embedded
-
- “sensitive tasks” that only a few dare to touch
-
- customers hear different versions from different people
-
- errors are fixed and forgotten, not captured
That’s not learning. That’s surviving on goodwill.
## The most common trap: the “dead manual”
Companies try to solve this with a huge document. It becomes a manual nobody reads, nobody updates, and eventually it’s dangerous because it creates false certainty.
A dead manual has three signals:
1) no owner
2) no expiry
3) no single searchable source of truth
If it doesn’t live in the workflow, it doesn’t live.
A living playbook isn’t a document. It’s a governed system.
I define it by properties:
-
- single source of truth (no five versions of the same rule)
-
- searchable in seconds (otherwise people will ask, and dependency returns)
-
- organized around decisions and events, not departments
-
- trace + expiry (without history there’s no learning; without expiry knowledge rots)
-
- minimal maintenance ritual (small, not bureaucratic—just enough to keep it alive)
## The most underestimated piece: a decision log
If I had to pick one thing—without “implementing” a big program—it would be a decision log.
Not a long report. A reconstructible trace:
-
- what we decided
-
- why
-
- under what boundaries
-
- what changes as a result
Because many frictions aren’t “how-to” problems. They’re “what is current” problems.
## How I test whether your knowledge is defensible (without tools talk)
I run behavior tests:
### The one-minute test
“Find the official answer in under 60 seconds.”
If it depends on asking someone, it’s not defensible.
### The handover test
“If the person changes, does the decision stay or get reinvented?”
Reinvention means the standard isn’t accessible.
### The exception test
“Where is it written which exception is allowed—and who approves it?”
If it doesn’t exist, exceptions will become the rule through fatigue.
### The update test
“How do I know what’s valid today?”
If no one can answer clearly, the system is fragile.
## Manuals that truly protect margin include boundaries
A playbook that only says “do this” doesn’t protect you.
The one that protects you also says:
-
- don’t do it if…
-
- if X happens, escalate to…
-
- if the data is missing, don’t promise…
-
- if there’s conflict, this rule wins…
That’s not “free consulting.” That’s a decision frame that prevents day-to-day erosion.
## The invisible risk: when knowledge becomes politics
In some companies, “knowing” equals power. Then documenting feels like losing influence.
If documenting threatens status, the company is designed for internal competition—expensive competition.
## Onboarding shows knowledge reality fast
If onboarding takes months to reach basic productivity, it’s rarely the new person’s fault.
It’s usually: scattered knowledge, implicit standards, and dependency on questions.
I don’t chase “perfect onboarding.” I chase predictable onboarding.
## The decisions I make before “writing more”
Before producing more content, I decide:
-
- what is critical (impact on promise, margin, quality)
-
- what format reduces friction (not what looks nicest)
-
- who owns each knowledge block
-
- what minimal ritual keeps it alive
Without these, documenting becomes accumulation.
## Closing
I don’t want a company with “lots of documents.” I want a company that remembers: it can reconstruct decisions, transmit criteria, and learn without repeating the same failures.
If you want, we’ll review it in 15 minutes: where you depend on memory today, what knowledge is trapped in people, and what minimal structure would give you continuity without bureaucracy.
