Abrigo·Marrón Systems

Apuntes de Campo Field Note

Teams Fail at the Seams

Two groups on a project stop cooperating. Each one's account of the problem is that the other group is unreasonable, and each account is delivered with real conviction and a list of examples. Management diagnoses a relationship problem, which is what it looks like, and books a facilitated session with an outside facilitator and a conference room with a flip chart.

You get two pleasant days, some genuine goodwill, and a laminated set of team norms. Within a month the friction is back, usually sharper, because now both groups have additionally learned that raising the problem through the official channel produces a workshop. The money bought nothing structural, because nothing structural was ever the subject.

The failure is almost always at the seam.

Two different functions, two different failures

The governance literature separates these cleanly, and the separation does real work. Operational control directs the productive units so the system holds its performance. Coordination keeps those units from interfering with one another. Different function, different information, different failure mode.

A control failure looks like a group not doing its job: missed output, wrong output, no output. It is visible, it has an owner, and organizations are generally competent at finding it.

A coordination failure looks like two groups who do not get along. Both are doing their jobs. Both are, by their own metrics, performing. The damage happens in the space between them, which belongs to neither and shows up on nobody's scorecard until the schedule slips and the inquiry begins. And because the failure has no owner, it gets attributed to the only thing visible at the boundary: the people standing on either side of it.

Those people are not the defect. They are where the defect becomes audible.

What the seam actually looked like

In the case that prompted this note — details generalized — the two groups were engineering and a maintenance planning organization, and the dispute presented as a personality conflict between two leads who had a long history of irritating each other. That history was real, and it was not the mechanism.

The mechanism was a two-week assumption. Engineering's schedule logic assumed that a design package handed to planning would come back reviewed in two weeks. Planning had never agreed to two weeks, had never been asked, and was carrying a turnaround backlog that made three to four weeks the honest number. Nobody had written the interface down, so both parties were operating correctly inside incompatible models of the same handoff.

Every two weeks, engineering experienced planning as obstructive. Every two weeks, planning experienced engineering as entitled. The interpersonal evidence accumulated on both sides for eighteen months and was entirely genuine — each group had a file of real incidents proving the other was difficult. The file was a symptom. The undefined interface was the disease.

It was repaired in a forty-minute conversation that produced a single page: what gets handed over, in what condition, what the committed turnaround is, who is notified when it will slip, and what engineering does with the slack in the meantime. The leads did not become friends. They stopped needing to be.

Make the handoff explicit and the interpersonal friction frequently evaporates. It was never about trust. It was about two competent people being set against each other by a seam nobody had drawn.

How to find the seams before they find you

Four questions at every boundary where one group's output becomes another group's input. They are boring questions and they are almost never all answered.

Two more diagnostics worth running while you are there.

Find the person holding it together informally. Every organization has one or two people who make things work by personally carrying information across a boundary nobody designed — the planner who calls the engineer directly because the official route is useless, the operator everyone phones before they phone the control room. They are usually the most valued people in the building, and they are also a precise map of where the designed system has a hole. Their heroism is the organization's unbudgeted maintenance cost on a defect it has chosen not to fix. When they retire, the hole becomes visible all at once.

Look for the function performed twice. Where two groups each believe they own the same responsibility, you do not have redundancy — you have a standing territorial dispute that will be read as personality for as long as it goes undiagnosed. Two parties performing one function in competition is a recognized governance deficiency, and it is resolved by assignment, not by mediation.

And when it really is the person

Sometimes it is. Occasionally someone is genuinely corrosive, and no amount of interface definition will fix a person who is choosing to be difficult. The order of operations still holds, because it is what lets you tell the difference.

Draw the seam first. If the friction resolves, it was structural, and you have fixed it cheaply and permanently without anyone having to be the villain. If the friction persists against a clean, written, mutually agreed interface, then you have something you did not have before: evidence about a person rather than an impression of them. That is a conversation worth having, and now it can be had on the merits.

Structure first is not softness toward bad behavior. It is the only sequence that produces a defensible conclusion about it.

The coordination and operational-control distinction follows the complex system governance reference model; see El Enfoque. The case is de-identified and composite, with timings and organizational units altered.

← Todos los apuntes