The short version
- A knowledge silo is not a storage problem. It is four people who are the only route to something, and a wiki does not change the route.
- Teams have usually tried a wiki twice already. The second attempt fails the same way as the first, for the same reason.
- Silos form along the seams of your org: by zone, by function, by tenure. Map them before trying to dissolve them.
- The measure of success is not pages written. It is whether a question can be answered without the silo being available.
Knowledge silos are usually described as an information storage problem, which leads directly to a wiki project, which is why most companies have two abandoned wikis and the same silos they had before.
"Everything lives in four people's heads."
What a silo actually is
A silo is a routing condition, not a storage condition. It exists when the only path to a piece of knowledge runs through a specific person or group. The knowledge may even be written down somewhere; if the reliable way to obtain it is to ask Ana, Ana is the silo.
That distinction changes what you do about it. Storage problems are solved by writing more. Routing problems are solved by creating a second path, and a document only creates a second path if people actually use it, which is an empirical question rather than an architectural one.
Why the wiki fails, twice
The first wiki fails on maintenance. Pages are written during a burst of enthusiasm, go stale within two quarters, and a reader who finds one wrong thing stops trusting the rest. Trust is the whole asset, and it is destroyed by a single confidently wrong page.
The second wiki fails on retrieval, which is subtler. This time the team is disciplined: there are owners, review dates, and a structure. And people still ask Ana, because finding the right page requires knowing what the answer is called. The person with a question does not know the vocabulary of the answer. Ana does.
Both failures come from the same source: the wiki asks the reader to do the work of locating the answer, and the reader has a colleague who will do that work for them in eleven seconds.
Map your silos in an hour
Do this with the team in one session. List the eight to ten areas the team depends on. For each, write the names of everyone who could answer a hard question about it without research.
| Names in the cell | What it tells you |
|---|---|
| One | A silo, and a genuine delivery risk |
| Two, in the same time zone | A silo for two thirds of the day |
| Two or three, across zones | Healthy for most areas |
| Everyone | Either genuinely shared or nobody is being honest |
The second row is the one that surprises distributed teams. Two people who can answer is comfortable until you notice both are in Amsterdam, which means the area has no coverage for the Asian working day. Silos on distributed teams form along time zones as readily as along expertise.
What actually dissolves one
Three things work, in increasing order of effect.
Pair across the seam. Put a second person from another zone on the area for a month. Expensive, slow, and it genuinely transfers judgement rather than facts.
Record decisions as they are made. Much of what appears to be deep expertise is a series of past decisions whose reasoning was never written. Four lines per decision removes a surprising amount of the silo.
Make the knowledge answerable without the person. This is the change that addresses the routing problem directly. With StandIn, each person confirms a ninety-second brief at the end of their day, and while they are off, their StandIn answers questions from it in their words, with a source under every answer, and never guesses. The second path to the knowledge exists, it is as fast as asking, and it does not depend on the reader knowing what the answer is called.
None of the three makes the expertise disappear, and that is not the goal. The goal is that the team is not blocked when the expert is asleep. See StandIn for leadership teams.
How to know it worked
Not by counting pages. Two measures are worth having. First, re-run the mapping exercise a quarter later and see whether any cell gained a name or a zone. Second, and more directly: pick an area with one name, have that person take a week off, and count how many times the team was blocked on them.
That second test is uncomfortable and it is the only one that answers the question. A silo that has been documented but still stops work when the person is away has not been dissolved, it has been described.
Common Questions
How do you break down knowledge silos on a remote team?
Create a second path to the knowledge that is as fast as asking the person. Pairing across zones transfers judgement, recording decisions captures reasoning, and making the record answerable removes the dependency on the person being awake.
Why do wikis fail at this?
Because they solve storage and not retrieval. A wiki requires the reader to know what the answer is called in order to find it, and someone with a question generally does not. So they ask a colleague, and the silo persists alongside the documentation.
Is some siloing healthy?
Specialisation is healthy and necessary. What is unhealthy is a single route: one person who must be available for an area to function. Aim for two or three people across at least two zones per critical area, rather than for everyone knowing everything.
What is the fastest thing we can do this week?
The mapping exercise, which takes an hour and usually reveals one or two areas nobody realised had a single name against them. That finding alone tends to change how the next month of work is allocated.
When you're off, your StandIn is on.
It answers your teammates' questions from work you've already done, in your words, with a source under every answer.