Who has worked with whom
Why the measure survives disagreement
Most team measures record opinion and move with the mood. A count of pairs records events, so two people who disagree still agree about the number.
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 a team wants evidence without reducing people to a score, managers can see how it fits and connect time and project patterns to a transparent discussion about workload.
An independent source worth comparing with this example is NCBI.
Ask a team whether it is working well and you will get eight answers shaped by eight different weeks. Ask them which pairs among them have done a piece of work together in the last quarter and you will get substantial agreement, including from the people who are unhappy.
That property — surviving disagreement — is the main thing the measure has going for it, and it is worth setting out why it holds and where it stops holding.
What a survey measures
Engagement and cohesion instruments ask people to rate statements. Whether they feel supported. Whether they can rely on colleagues. Whether they would recommend the team. The answers are real data about real experience and they have a specific set of properties.
- They move with recent events. A difficult fortnight moves every score on the form, including the ones about things that have not changed.
- They move with who is asking. Everybody knows the form is read by somebody, and most people adjust, usually without deciding to.
- They cannot be checked. If two people give opposite answers, both are correct. There is no fact either of them is wrong about.
- And they cannot distinguish the structural from the atmospheric. A team that is unhappy because it is a wheel and a team that is unhappy because of one difficult relationship produce similar scores and need completely different responses.
What a pair count measures
An event, or its absence. Either these two people produced something together inside this period or they did not. It is a question about the past with an answer, and the answer is held by the two people involved.
That gives it three properties the survey lacks.
It is checkable. If two people disagree about whether a pair exists, one of them is remembering wrongly or the bar was ambiguous, and both of those can be resolved by looking.
It is stable under mood. A bad month does not remove a pair that happened. This is why the count is useful precisely in the weeks when the mood instruments are least informative.
It is specific. The output is not a score but a picture with named gaps, and each gap suggests a particular action involving two particular people.
Where it stops holding
The measure is narrow, and three limitations are worth stating on a page that is arguing for it.
It records that work happened, not whether it was good, whether either person wants to repeat it, or whether it should have needed two people. A pair formed by a painful handover counts the same as a pair formed by a productive afternoon. If something has gone wrong between two people, the grid will show a healthy pair.
It records a period and goes stale. A count from a year ago describes a team that no longer exists, and the pairs in it may have decayed to the point where only the goodwill survives.
It depends entirely on the bar. Two teams with different definitions of "worked together" cannot be compared at all, and the temptation to compare is strong the moment two numbers exist.
The property that makes it fragile
Everything above holds only while nobody is rewarded for the number.
A measure that records events survives disagreement because the events happened and both parties were there. It stops recording events the moment there is a reason to produce them: pairs get arranged so that the grid can show them, two people put their names on one thing they did separately, and the bar quietly relaxes because everybody involved would prefer a better figure.
This is not a risk peculiar to this method; it is what happens to any descriptive measure that is promoted to a target. It is worth stating here because the whole argument of this note — that the number is checkable and stable — is exactly what is lost first.
The practical consequence
Use both, for different questions, and do not substitute one for the other.
When the question is how people feel, ask them. There is no structural proxy for whether somebody is miserable, and inferring it from a grid is worse than asking.
When the question is what the team is shaped like and what to change, count. Opinions about shape are unreliable in a predictable direction: everybody overestimates the connectedness of the part of the team they are in and has no information about the rest.
The most useful case is when the two disagree. A team that reports feeling connected and has a sparse grid is usually a team where each person is well connected to the manager and reads that as connection to the team. A team that reports feeling disconnected with a dense grid usually has a problem that is not structural at all, and spending a quarter on pairings will not touch it.
Neither of those is visible from one instrument alone.