The matrix organization solved a real problem and created a quieter one. By letting a person serve more than one line of work, a functional manager and a project lead, say, the matrix gives companies flexibility to staff across boundaries. The quieter problem is what happens to the person sitting at the intersection. They have two managers, and each of those managers, when planning the quarter, mentally allocates the person as if they were fully available. Neither manager can see the commitment the other one made, so both plan against the same hours, and the same hours cannot be in two plans at once.
The engineer at the center knows about the conflict before either manager does, because the conflict lands on them first. They are the one holding two sets of expectations that sum to more than a week, and they resolve it the only way an individual can: by silently working extra hours, by letting one side slip, or by context-switching so heavily that both sides get a degraded version of their attention. The system's over-allocation becomes the individual's stress, and it stays invisible to the people who created it because each manager only sees their own half.
The conflict is structural, not interpersonal
It is tempting to frame this as a communication failure. The managers should talk. The engineer should speak up. And yes, those things help at the margin. But the root cause is that the two claims on the person's time exist in two separate plans with no shared view, so neither manager has any way to see that their 100% and the other manager's 100% add up to 200% of a person who has 100% to give. Each plan is locally reasonable. The conflict only exists in the sum, and nothing computes the sum.
This is the accountability gap in its capacity form. Each manager is accountable for their own deliverables and reasonably assumes the resource is there. No one is accountable for the total load on the shared person, because the total load is not visible anywhere. The person is over-committed by the structure, and structural over-commitment cannot be talked away by better intentions, because the next quarter's planning will reproduce it exactly as long as the two plans stay blind to each other.
Why it stays hidden until it breaks
The over-allocation is invisible right up until it produces a failure, and the failure is usually misattributed. A deliverable slips, and the manager whose work slipped sees a performance problem with the engineer, not a capacity problem they helped create. The engineer, who has been absorbing the conflict through unsustainable hours, eventually burns out or disengages, and that too reads as an individual issue rather than a structural one. The matrix quietly manufactures these outcomes and then blames the people standing where its seams meet.
This is the human cost of treating capacity as something other than a first-class strategic input. When the company plans work without a shared, honest model of who is committed to what, it routinely allocates more than 100% of its people and discovers the over-allocation only through slipped deadlines and attrition. The shared engineer is just the most acute instance, the place where the company's habit of over-committing capacity becomes one specific person's problem.
A shared view is the whole fix
The resolution is not more coordination meetings. It is a single, shared view of each person's commitments across every line of work they serve, so that when a second manager goes to allocate the engineer, the existing claim from the first manager is already visible, and the new commitment lands against the remaining capacity rather than against an imaginary full week. The conflict gets surfaced at planning time, to the managers, instead of at delivery time, to the engineer.
This is what it means to give work explicit owners and visible load rather than letting commitments accumulate in separate plans. When a person's total allocation is something you can see, over-commitment becomes a decision someone has to actively make and own, instead of an accident that emerges from two reasonable plans that never met. The managers can then negotiate the trade-off directly, which is a conversation they can only have if they can both see that the trade-off exists. Right now, in most matrix orgs, the only person who can see it is the one with no authority to resolve it.
What to do this quarter
Find the people in your org who report into more than one line of work, and for each one, try to reconstruct their total committed load across all their managers. The exercise is harder than it should be, which is itself the finding: if you cannot easily see one person's full allocation, neither can the managers committing them, which means you are almost certainly over-allocating them without knowing.
Then make the shared load visible for at least the most matrixed roles. The goal is that no manager can commit a shared person without seeing what that person is already committed to. It does not require a heavy process. It requires that the claims on a person's time live in one view instead of several, so the sum is computable by someone other than the exhausted individual currently computing it in their head at 11pm.
FAQ
Why does a shared engineer end up over-allocated? Because each of their managers plans as if they have the person full-time, and neither can see the other's claim. The two commitments live in separate plans, so nothing computes that they sum to more than one person's capacity. The conflict only exists in the total, and the total is not visible to anyone except the engineer absorbing it.
Isn't this just a communication problem? Better communication helps at the margin, but the root cause is structural: two claims on the same hours in two blind plans. Next quarter's planning reproduces the over-commitment regardless of intentions, because neither manager can see that their 100% plus the other's 100% exceeds what the person has. It is the accountability gap in capacity form.
Why does the over-allocation stay hidden? Because it only surfaces as a slipped deliverable or burnout, both of which get misattributed to the individual rather than to the structure that over-committed them. The matrix manufactures these failures at its seams and then blames the people standing there, which is what happens when capacity is not a first-class input.
What actually fixes it? A single shared view of each person's commitments across every line of work, so a second manager allocating the engineer sees the existing claim and commits against remaining capacity. Giving work explicit owners and visible load turns over-commitment into a decision someone must own, surfaced to managers at planning time rather than to the engineer at delivery time.