Who is still here
Part-time members and borrowed people
A designer shared across three teams is on three grids and belongs to none of them. How to count somebody whose attention is divided, and what they need instead.
The people this page is about, and which of them have ever worked together
- 01The team's managerTwo of this person's five working pairs
- 02The senior engineerOn the team from the first day
- 03The second engineer
- 04The product managerThe third person in this role in fourteen months
four people · 6 possible pairs · 3 of them have worked on something together · 1 has only been in the same meeting · 2 have never done either
Four of the eight, and only the pairs that run between those four.
Before introducing another status meeting, it helps to learn about the available tools and decide whether a clearer view of time and projects would answer the question first.
Teams can compare this approach with the practical guidance available from Holacracy.
Two of the eight people counted on this site are not really eight people's worth of person. The designer is shared with two other teams. The data analyst reports here and spends most of their time answering questions from a different division.
Both appear on the roster at full width. Both are counted in the denominator — twenty-eight pairs assumes eight whole members — and neither can plausibly maintain seven working relationships here. The grid overstates what is available from them, and this note is about what to do about that.
Count them, and mark them
The alternative is to leave them off, and that is worse. They do real work with real people on this team, those pairs exist, and a grid without them describes a team that does not exist either.
So they go on the roster with a note. The note matters because it changes what the empty cells in their row mean. A full-time member with two pairs out of seven is under-connected. A one-day-a-week member with two pairs out of seven is doing about as well as the arrangement permits.
What the note should say is what to expect from the row, which is a different figure for each kind of membership:
| On the roster as | Realistic pairs here | What to plan for |
|---|---|---|
| A full member | Five to seven | The ordinary mechanisms |
| Shared with two other teams | One, perhaps two | One deliberate pair, at the seam |
| On loan for a quarter | None | Separable work and one contact |
What a shared person actually has
Not a fraction of eight relationships. Something closer to a fixed small number, regardless of how many teams they are on.
Maintaining a working relationship costs attention that does not scale with how many teams you belong to. Somebody on three teams does not get three times the capacity; they get the same capacity divided three ways, and the usual result is one or two real pairs on each team and a great many acquaintances.
So the realistic expectation for a shared member is one strong pair here, with whoever their work most touches, and nothing else. Planning for more produces a person spread so thin that none of the pairs is load-bearing.
Choosing the one pair
This is the useful decision and it is almost never made deliberately.
By default, the one pair a shared person forms is with whoever asks them for things most often, which is usually the manager or whoever is loudest. That is an accident of volume.
Chosen on purpose, it should be with whoever is on the other side of the seam their work crosses. For a shared designer, that is whichever engineer builds what they design. For an analyst, whoever consumes their numbers. One deliberate pairing at that point is worth more than five scattered acquaintances, and it is the pair that fails expensively when it does not exist.
The rituals problem
Shared members are the people for whom whole-team rituals cost the most. A weekly meeting is a tenth of their week here. Two such meetings plus a social is most of their day.
The consequence is a reliable and unfair pattern: the people with the least time on the team spend the highest proportion of it in the activities that produce nothing, and the lowest proportion in the work that would produce a pair.
The fix is uncomfortable and simple: excuse them from the ceremonies and include them in the work. A shared designer who skips the weekly meeting and reviews one engineer's work every fortnight is more connected to the team than one who attends everything and builds nothing with anybody.
The borrowed person
Worse than shared is borrowed: somebody on loan for a quarter, counted in the headcount, named in the plan, and gone before any pair has formed.
By the measure used here they are a person with a near-certainty of an empty row, because the time to a first working pair is about a quarter and so is the loan.
The honest handling is to not expect a pair and to plan the work accordingly. A borrowed person should be given something separable, with one named point of contact for everything, and nobody should be surprised when their departure changes nothing about the network. The mistake is to treat the loan as team growth and to be puzzled a quarter later that the team feels the same size.
The denominator question
There is an argument for not counting shared people in the denominator at all — twenty-eight pairs assumes eight whole members, and two of these eight are not whole.
It is tempting and it should be resisted, for the same reason the sign on this site refuses to draw two different team sizes on one pyramid. The moment the denominator becomes a judgement, the number stops being comparable with itself: next quarter somebody will weigh the designer differently, and the change in the count will be a change in the arithmetic.
Keep the denominator crude and put the nuance in the notes. A number that is slightly wrong in a fixed way is more useful over four quarters than one that is adjusted to be fair each time.
On the roster note
Write the note next to the name, not in a footnote. The whole value of marking a shared member is that the next person reading the grid — which may be you, in six months, or a new manager — adjusts their expectation before they interpret the empty cells rather than after.