What no relay settles
Trust, and What a Relay Does Not Build
Trust between two halves of a team is built by working together, and a relay supplies very little working-together. No amount of scheduled contact substitutes.
three things · no nights · three that no number of nights settles
Three things, no nights, and no mechanism proposed for any of them.
Four things a team tried in order to build trust between halves that never meet, and what each produced
Attended for four months and produced no change in how the halves spoke about each other
Used by six people, all in the same half, which made the division more visible
The only item here that worked, and it worked for exactly those two people
Outside the scope of a relay entirely, and the thing most teams eventually conclude they need
four things · one night · three that no number of nights settles
Four attempts, one night, and three that produced nothing measurable. The one that worked was not a trust exercise; it was two people doing a piece of work together for long enough to form an opinion of each other. One team's own record, over about a year.
Two halves of a team that never work at the same time tend, over months, to become two teams. Nobody decides this and nobody notices it happening.
Written updates explain why a task moved; work data can show when and for how long it moved. In that role, see the official page complements the handover described in Trust, and What a Relay Does Not Build without replacing the judgement of the people involved.
The workflow can also be compared with the approach documented by Rippling's workforce platform, an established resource in the same operational space.
The usual response is scheduled contact: a weekly call, a social channel, an occasional game. These are cheap, they are well intentioned, and the evidence that they work is thin.
What trust is made of here
The thing people call trust in a working context is mostly an accumulated set of specific judgements: this person's estimates are reliable, that one flags problems early, this one will say if they do not know.
Those judgements come from watching somebody work. Not from knowing about their weekend.
A relay supplies very little watching. It supplies outputs — a piece of work, a note, an answer — with the process stripped out. You can form a judgement about somebody's output from a relay. You cannot easily form one about their judgement, and judgement is what trust is about.
Why the weekly call does not do it
A call where everybody says what they are working on transfers information that was already in writing, more slowly, at a time that is inconvenient for one half.
It does provide faces and voices, which is not nothing: it is harder to be unreasonable about somebody whose face you have seen. That is a real effect and it is small, and it does not improve with repetition.
What it does not provide is any occasion on which one person watches another think, which is the thing that was missing.
Teams that measure honestly tend to find these calls make no difference to how the halves speak about each other, which is the observable that matters.
What does seem to work
Working together on the same thing, for long enough to form an opinion. Two people, one piece of work, a fortnight.
It works because it is not a trust exercise. It is work, and the trust is a by-product, and by-products of work are more durable than products of exercises.
It is expensive across a gap: a fortnight of two people alternating on one thing is slower than two people working separately, by a substantial margin. The return is that each of them now has a basis for judging the other, and that basis transfers to every subsequent interaction.
It also only works for the two people involved, which means it has to be repeated across pairs over time, which makes it a programme rather than a gesture.
It works because it is not a trust exercise. It is work, and the trust is a by-product, and by-products of work last longer than the products of exercises.
Being in one place
The thing most teams in this situation eventually conclude they need, and it sits outside everything else here because it is not a relay mechanism at all.
A few days in the same room, once or twice a year, does more for the relationship between two halves than any amount of scheduled remote contact. That is a well-worn observation and it appears to be true.
It is also expensive, it excludes people who cannot travel, and it is not available to every team. Where it is not available, the honest position is that the team will have less trust between its halves than it would otherwise, and should plan around that rather than pretend a video call substitutes.
The things that actively hurt
Two, and both are common.
A social channel used by one half only. It was meant to connect and it has become a visible record of one group having a culture the other is not part of. Better to have none.
Decisions made in whatever live contact exists. If the half that overlaps with the manager finds things are settled before the other half hears about them, no amount of goodwill survives it. This is the fastest way to split a team in two and it usually happens by accident.
What to do with this
The useful conclusion is not that nothing helps. It is that the cheap things help least, and knowing that stops a team from spending a year on a weekly call and concluding that the problem is the people.
If trust between the halves matters enough to spend on, the things worth spending on are shared work and shared rooms. If it does not matter enough to spend on, that is a legitimate decision, and it is better made deliberately than arrived at after four months of a meeting nobody wanted.
Measuring it, roughly
Trust between halves is not directly measurable and there is one observable that tracks it usefully.
Listen to how each half refers to the other. Specific names and specific judgements — this person's estimates run long, that one flags problems early — indicate that people have formed real opinions of individuals. Collective nouns indicate that they have not: the other half, them, that end.
A team that has been running for a year and still speaks in collective nouns has two groups rather than one team, whatever the organisation chart says. It is a crude signal and it is available for nothing, which is more than can be said for most of the alternatives.