A matrix is a deliberate trade: better coverage of two dimensions, paid for in decision speed and in ambiguity about who settles what.
Why it matters when the plan changes
Matrices fail in a predictable way. Consultation improves, because more perspectives are structurally present. Authority degrades, because more parties have standing and none has primacy. The responsibility assignment matrix convention assumes accountability is individual and final, which is exactly what a matrix leaves unassigned. Where the design does not say what happens when the two lines disagree, escalation answers the question instead, and escalation becomes the operating mode rather than the exception.
The tension is that the ambiguity is often the point. Leaders adopt a matrix precisely because they do not want to choose between two priorities, and the structure defers the choice into thousands of individual interactions. The organisation then makes the trade-off repeatedly, inconsistently, at a level with no view of the whole. Naming a single owner for the recurring decision usually resolves this; redesigning the whole structure rarely does.
In practice
A product manager reports to a product line and a regional business. The two disagree on launch sequencing. The design says both must be consulted and does not say who decides. The manager escalates, twice, and the decision is eventually made by a chief executive who has the least context of anyone involved.
Evidence
Matrix management gives individuals more than one reporting line, typically across function and product or project.
Matrix management, Wikipedia (2026)The responsibility convention treats accountability as individual and final, which is what a matrix most often leaves unassigned.
Responsibility assignment matrix (RACI), Wikipedia (2026)
What it cannot tell you
A matrix organisation describes reporting structure, not decision rights. It cannot tell you who has final say when the two lines disagree, whether an escalation route exists, or how disputes actually get resolved. Two organisations with identical matrix diagrams can differ entirely in speed and clarity, depending on rules the structure itself never specifies.
Questions
When an organisation genuinely has to optimise for two dimensions at once and the cost of choosing one is higher than the cost of coordinating both. It works where the tie-break rules are explicit and fails where the ambiguity was the reason for adopting it.
Written rules for who decides what, and what happens when the two lines disagree, settled in design rather than in the moment. One named accountable owner per decision, a short consulted list, and an escalation route that is used rarely because it is rarely needed.
Because when two parties have standing and neither has primacy, escalation is the only available resolution. Each individual escalation is reasonable. In aggregate, though, the decisions migrate upward to a level that has the least context and the least available time.
It slows the ones it fails to assign and has little effect on the rest. That is why the cost is uneven and hard to see: most decisions are unaffected, and the small number that cross the unassigned boundary consume a disproportionate share of senior attention.
Usually by assigning rather than redesigning. Listing the decisions that recur at the boundary and naming a single owner for each resolves most of the observed friction, and it costs a short series of conversations rather than another full restructure of the organisation.