Tribal knowledge isn't quaint. It's an operational risk.

When you describe your company to a new hire, you probably tell them to "ask Marco about anything client-facing" and "ask Anna for anything operations". You may even say it with a touch of pride. It is the kind of detail that signals a tight, capable team where people know each other and know who to go to.

It is also a description of two single points of failure.

The framing matters. "Tribal knowledge" sounds folksy. It sounds like a feature of small-company culture. It is not. It is exactly the same class of problem that, in any other context, you would call concentration risk. And it is one of the most underappreciated risks on the operational side of every SME we work with.

Why "ask Marco" feels good and is dangerous

Concentrating knowledge in a few people works extremely well for a while. It is efficient. Decisions get made quickly. The hand-trained people are very good at what they do. New problems route to the right person almost automatically.

The trouble is that the system has no resilience. If Marco gets sick for a week, the parts of the company he holds in his head become unreachable. If he gets recruited away, the parts of the company he holds in his head go to a competitor. If he becomes a bottleneck — which always happens, because anyone whose knowledge is centralizing becomes the queue — the entire company's throughput is capped at his.

You would not run your accounting this way. You would not let one person be the only one who knows where the bank account passwords are. You apply the principle elsewhere instinctively. It just happens to be invisible on the knowledge side, because nothing breaks until it suddenly does.

The three flavors of tribal knowledge

It helps to separate the kinds of knowledge that tend to concentrate.

Process tribal knowledge is the "how we actually do things around here". Which approval steps are real and which are theatre. What the unofficial workflow is when a client urgently needs something. Why we use Tool A for one client and Tool B for another. This is mostly recoverable, if anyone bothers to ask, but no one does until it is already a problem.

Relationship tribal knowledge is much harder. It is the patterns of trust and tone with specific contacts. Who responds well to formality, who hates it, who needs reassurance, who needs results. This kind of knowledge is built up over months and years of conversation. It is almost never written down. When the person who holds it leaves, the relationships visibly degrade.

History tribal knowledge is the "why we did it this way". The reasons behind decisions that, viewed in isolation today, look strange. The constraint you were under three years ago that no longer applies but that nobody dared to revisit. The compromise you made with one supplier that has been quietly costing you margin ever since. This is the most expensive to lose, because new employees, lacking the history, will either repeat the mistake of trying to change it (and being told off by the people who remember) or accept it as gospel forever.

Documentation does not fix this

The standard advice is to document tribal knowledge. Get the people who know things to write them down. Build runbooks. Use Notion.

If you have ever actually tried to make this happen, you know how it goes. The first attempt produces five well-written pages. The next quarter produces two more, of lower quality. By month six, the documentation effort has quietly stalled, and the pages that exist are partially out of date. Meanwhile, every day, more tribal knowledge is being accumulated through normal operations.

This is not a discipline problem. It is structural. The people who know things are busy doing the work that produces the knowledge. Asking them to also write a parallel document is asking for time they do not have. And when they do find the time, they write down what they can articulate, which is a small fraction of what they actually know.

The conversational layer is already capturing most of it

Here is the part that gets overlooked. Almost all the tribal knowledge in your company is already being expressed, just not in documents. It is in:

• Slack threads where Marco patiently explains, for the fourth time this year, why we do something a certain way.
• Email replies where Anna walks through the reasoning on a procurement decision.
• Meeting transcripts where someone offhandedly mentions a constraint that everyone else takes as given.

If you could read across these surfaces, you would find that the tribal knowledge has been articulated, repeatedly, over months. It just was not collected.

The fix is not to ask Marco to write it down. The fix is to make readable the conversations where he has already explained it, hundreds of times, to other people. The "ask Marco" pattern would still exist, but Marco would no longer be the only one who could answer.

The risk version of the conversation

When you reframe tribal knowledge as concentration risk, the people in your organization who care about risk start caring about it. Your CFO will not get excited about better documentation. Your CFO will get extremely interested in the question "what is our operational exposure if any one of our five key people leaves in the next six months?"

The answer, at most SMEs, is something between "alarming" and "catastrophic". The exposure is not on any spreadsheet. It would be — visibly — if you reframed it.

Once it is on the spreadsheet, the conversation about how to reduce it becomes a conversation about risk reduction, not about culture. And risk reduction has budget. Culture initiatives almost never do.

What "de-risking" tribal knowledge looks like

Concretely, two things change at companies that take this seriously.

The first is that the conversations where decisions get made and explained — the Slack threads, the email exchanges, the meeting recordings — stop being treated as ephemeral. They become structured, searchable, and consolidated into a layer that anyone (or any AI agent) in the company can read on demand. The "ask Marco" reflex is replaced with "ask the system, and the system shows you what Marco has said about this in the past two years".

The second is that the hiring of senior people stops being so frightening. The fear of senior turnover comes from the gap they would leave. If the gap is mostly closeable from what they have already said, the fear drops by an order of magnitude.

Tribal knowledge is not going away

You are never going to fully eliminate the pattern of certain people knowing more than others. You should not try. Concentration of expertise is, in moderation, a good thing.

What you can change is the cost of that concentration. Today, "ask Marco" means "if Marco is unavailable, you are stuck". Tomorrow, it can mean "Marco has answered some version of this thirty-six times over the past two years, and the answers are right here". That is not a culture change. It is a risk position change. And it is by far the cheapest way to make your SME more resilient without losing what made it efficient in the first place.