Who has worked with whom
What the count looks like after a year
Four counts a year produce a shape rather than a number. What moves on its own, what only moves when somebody causes it, and how to read a count that went down.
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
When the manager is the only route between people, the group can review this product overview and use project-level patterns to identify where direct collaboration would remove delay.
The broader organisational question can also be checked against resources from National Institutes of Health.
One count is a baseline and not much else. Four of them, taken the same way at the same intervals, start to show which parts of the shape are weather and which are climate.
This note is about reading the sequence, including the two results that look like failures and are not.
What moves on its own
The total drifts with how much work there is. A quarter with three large pieces of work produces more pairs than a quarter with twelve small ones, because large pieces need more than one person and small ones do not. That drift is not a fact about the team's health and should not be read as one.
The total also moves with arrivals and departures, and it moves asymmetrically. Somebody leaving removes every pair they were in at once. Somebody arriving adds zero pairs and seven possible ones. A team that replaced two people in a quarter will show a drop that has nothing to do with anything anybody did.
Near-misses — pairs that have met but not worked — accumulate on their own, because meetings happen without being arranged. A grid that has filled up with the middle category and not the first one has been busy and has not changed.
What only moves when somebody causes it
Pairs that do not include the manager. This figure is almost inert under normal operation, because the default routing of work creates hub pairs. If it moved, something was done on purpose, and it is worth knowing what.
Isolated people becoming connected. Nothing on a normal quarter reliably gives a pair to somebody who has none; the mechanisms that would have done it are the ones they are already outside. This moves when somebody arranged it.
Pairs between the clusters. If a team contains two sub-groups, work inside each one produces pairs steadily and work between them produces none, because there is none. A bridge pair appearing is always the result of a decision.
Those three are the honest success measures for anything tried. The total is not.
Reading a count that went down
It happens, and the first instinct is to treat it as a regression. Usually it is one of four things, and only the last is bad news.
Somebody left. Check how many of the lost pairs were theirs before concluding anything. A hub departure can halve a grid in one week.
The work changed shape. A quarter of individual maintenance work after a quarter of joint building will produce fewer pairs and may have been exactly the right thing to do.
The bar tightened. If the person counting has become more rigorous — and they usually do, between the first count and the second — the drop is in the instrument, not the team. This is the commonest cause and the easiest to miss, which is why the rule is written at the top of the grid.
Or the pairs genuinely decayed: the people who used to work together were given separate things and have not spoken since. That is the real finding, and it is almost always a consequence of a staffing decision taken for unrelated reasons three months earlier.
What a good year looks like
Not a rising total. A team whose count rises every quarter for a year is either growing, lowering its bar, or manufacturing collaboration.
A good year looks like a total that stays roughly flat while the distribution changes: fewer pairs concentrated on one person, no isolated members for two consecutive quarters, and at least one pair spanning whatever the team's main internal seam is. The team is doing the same amount of joint work and it is arranged better.
The second thing a good year produces is a boring answer to a question that used to be contentious. Somebody asks whether the team is really a team, and instead of an argument there is a grid and a date. Everybody looks at it and agrees about what it shows, and then argues about what to do, which is the useful argument.
Keeping the four counts comparable
Three things have to stay fixed across the year or the sequence is four unrelated numbers.
- The rule, written at the top of the grid and not rewritten. If it has to change, say so and treat the next count as a new baseline rather than a continuation.
- The roster boundary. Decide once whether shared and borrowed people are in, and keep the answer. A team that quietly drops its part-time designer in the third quarter will show an improvement it did not make.
- And who adjudicates. If the first count was compiled by asking each person and the third was reconstructed by the manager from memory, the difference between them is mostly a difference of method.
When to stop counting
When the answer stops changing anything. If three consecutive counts have produced the same picture and no decision has been made differently because of them, the measurement has become a ritual, and this site is against rituals that produce no pairs — including its own.
Put it down and pick it up when something changes: a reorganisation, a new manager, a team that has grown past about ten. Those are the moments when everybody's intuition about the shape is wrong and a count is worth twenty minutes.