Who has worked with whom
Reading a sparse matrix without panic
Most first counts come out low and the first reaction is alarm. A guide to what the number actually indicates, what a normal range looks like, and when to worry.
The people this page is about, and which of them have ever worked together
- 01The team's managerThree of this person's five working pairs
- 02The senior engineer
- 03The second engineer
- 04The designerShared with two other teams
- 05The QA engineer
five people · 10 possible pairs · 4 of them have worked on something together · 2 have only been in the same meeting · 4 have never done either
Five of the eight, and only the pairs that run between those five.
Home / Who has worked with whom
For a distributed group, teams can read the full overview to review where work travels, where it stalls, and whether responsibility is concentrated in one person.
The broader organisational question can also be checked against resources from Britannica.
The first time a team is counted, the number is lower than everybody expected. Six out of twenty-eight. Four out of fifteen. Nine out of forty-five. The reaction in the room is some mixture of surprise and embarrassment, and somebody usually says that this cannot be right.
It is almost always right, and the expectation was wrong. This note is about calibrating the expectation.
Why a full grid is not the goal
A team where every pair had worked together would be a team that had passed every piece of work through every person. For eight people that is an enormous amount of coordination, and the work would have been slower, not better. Beyond a certain density, additional pairs are additional meetings.
There is also an upper bound set by arithmetic that nobody budgets for. Maintaining a working relationship costs something — attention, context, the occasional conversation that is not strictly necessary. If that cost is a couple of hours a month and a person has twenty working days in it, there is a ceiling on how many live relationships any one person can hold, and it is somewhere around five to seven. Eight people each holding six relationships would give twenty-four pairs out of twenty-eight, and nobody would have time to do anything.
A rough calibration
These are not benchmarks and there is no dataset behind them. They are the ranges that make sense given the arithmetic above, offered so that a first count has something to sit against.
For a team of eight, around a quarter of the possible pairs — six to eight — is sparse but functional if the right ones exist. Twelve to sixteen is a well-connected team. Above twenty, either the team is really smaller than eight or the bar was set too low.
The ranges scale badly with size, which is the point of the whole subject. A team of four with six of its six pairs is dense and normal. A team of twelve with six of its sixty-six is almost certainly several teams.
What to look at instead of the total
Four readings, all more actionable than the headline number.
- Isolated people. Anybody with no pair at all. This is the single most useful thing on the grid, because the fix is small and the effect is immediate: one task, one quarter, and that person has a row.
- Pairs that bypass the hub. How many relationships exist that do not include the manager. A hub-heavy grid has specific costs, and they are set out in their own note.
- The near-misses. Pairs that have met but never worked. These are the cheapest to convert, because the two people have already found out that the other exists and is reasonable. A near-miss is most of the way to a pair and needs a task, not an introduction.
- The seams. Go through the last quarter and find the places where the work actually crossed between two people — a handover, a decision that landed on somebody, a problem solved twice. Then check whether those pairs are marked. An unmarked seam is where the count turns into money.
When to worry
Worry about an isolated person who has been on the team for more than one quarter. Everything else on the grid is a description of how the work is arranged. This is a person who has been present at every ceremony and has produced nothing with anybody, and that situation decays: they become harder to involve the longer it lasts, and they are the likeliest to leave.
Worry about a grid where the only pairs run through one person and that person is leaving.
Worry about a count that went up a lot without anybody arranging it, because the usual explanation is that the bar moved.
Do not worry about a low total on its own. A low total on a team doing eight separate things is an accurate description of eight separate things.
Who should see it
The whole team, with roles rather than names, and at the same time.
A grid held privately by the manager becomes a document about individuals, because that is what a private list of people's relationships is. A grid shown to everybody becomes a description of how the work is arranged, which is what it was meant to be.
There is a second reason, and it is the stronger one. Every person on a sparse team privately believes the thinness is specific to them — they can see their own few connections and nobody else's. Showing the grid disposes of that in one minute. It is the only part of this exercise that reliably produces relief rather than work.
What the number is for
It is a baseline. Its value is almost entirely in the second measurement: the same team, the same rule, three months later, after you have tried something. Without the first count the second one is just another number that feels about right.
That is the whole claim made for the method here. It is not a diagnosis, it is not a score, and it does not say whether a team is good. It says what the shape was on a particular date, so that the next shape can be compared to it.