Most matrix organisation problems come down to one question nobody answered: who decides when the two lines disagree. Research among managers in matrix structures finds unclear roles, ambiguous authority and misaligned goals at the top of the list, and matrixed employees are less likely to know what is expected of them. Decisions slow where authority is shared without being assigned.
Where do matrix organisations usually slow down?
At the points where two reporting lines meet and neither holds the decision. A country head and a global product lead both have a legitimate claim on a pricing, hiring or investment call. When the design did not say who decides, the decision escalates, or gets made twice. The structure is not the problem. The unassigned decision is.
Sy and D'Annunzio asked 294 top-level and mid-level managers across seven multinational corporations about their matrix organisations and found five recurring challenges: misaligned goals, unclear roles and responsibilities, ambiguous authority, the lack of a matrix guardian and silo-focused employees. Three of the five are versions of the same gap. Goals are misaligned because each line sets its own. Roles are unclear because the design named the boxes and not the decisions. Authority is ambiguous because it was shared in principle and never assigned in practice.
Who feels the problem first, and why does the top team miss it?
Middle managers feel it first and most. In the same study, 87% of mid-level managers cited unclear roles and responsibilities as a major issue, against 23% of top-level managers. The top team designed the matrix and knows what it intended. The managers inside it live with what was actually assigned, which is less.
The gap between those two numbers is the diagnostic. A leadership team that asks itself whether the matrix is working will usually say yes, because from where it sits the lines are clear. The question has to be put to the people two levels down, at the intersections: for the last three decisions that stalled, who did you think owned them, and who did the other line think owned them? The answers rarely match, and the mismatch is the slowdown.
Do matrix organisations get anything in return?
Yes, and that is why they persist. McKinsey's analysis of Gallup data on nearly 4,000 US workers found that matrixed employees reported better collaboration than their non-matrixed peers, and at the same time were less likely to know what was expected of them. The trade is real. The question is whether the clarity side of it was ever designed.
In that research, 84% of the employees were matrixed to some extent, and only a minority of the most heavily matrixed strongly agreed that they knew what was expected of them at work, against 60% of those who were not matrixed. Collaboration is what a matrix is built to produce, and it tends to deliver it. Clarity is what a matrix consumes, and it has to be put back deliberately, decision by decision, or the organisation pays for the collaboration with speed.
Why does the matrix need an owner?
Because nobody inside either line has an incentive to resolve the conflicts between them. Sy and D'Annunzio named the lack of a matrix guardian as one of the five challenges: a role that monitors how the matrix performs, surfaces the decisions that stall and settles the disputes neither line will raise. Without it, conflicts are settled by escalation or attrition.
Davis and Lawrence catalogued nine pathologies to which the matrix is particularly vulnerable in 1978, and the list has not dated: power struggles between the lines and what they called decision strangulation among them. The modern version of the guardian role is often a chief operating officer or a head of the operating model with a short list of standing questions. Which decisions stalled this month? Which were taken twice? Which escalated to the executive team that should have been settled a level below? A matrix with that list, reviewed, behaves very differently from a matrix without it.
Can the slowdown be fixed without another restructuring?
Usually, and usually faster. The fix is to assign the decisions, not redraw the boxes: list the decisions that stalled or were taken twice, give each one owner, state what the other line is consulted on, and review the list quarterly. A matrix that stalls has usually skipped this step, not chosen the wrong structure.
The order matters. Start with the decisions that carry the plan: pricing, key hires, capital, product priorities in the markets that matter most. Settle those first and let the long tail wait. A business-area leader can do this for their own area in a quarter, without waiting for a group-wide operating model review, and the result is usually visible in the escalation list within two cycles. Where the same decision keeps stalling after it has an owner, the problem has moved from design to the person or the relationship, and that is a different conversation.
Where this stops
This note is about decision speed in a matrix that already exists. It does not say whether a matrix is the right design for a given business, how to weight regions against product lines, or how to run a full operating model review. The studies cited describe the challenges managers report across many companies and cannot say which decisions are stalling in a particular organisation, or why.
How Atlas reads it
Atlas treats a new operating model, including a matrix that is two years old and still being argued over, as one of the six moments that should prompt a reading of the organisation behind the plan. The reading sets what the plan demands of the structure against what the structure presently supports, and the gap usually shows up as decisions without an owner at the points where the lines cross.
The recommendation follows the intervention ladder: clarify which decisions carry the plan and who owns them first, resequence where the same people are the bottleneck for several lines, reassign the decisions that keep escalating, and reach questions about individual roles last. Each judgement carries the evidence for it and what argues against it, so the leadership team can see what a change to the design would actually trade.
Sources
- Thomas Sy and Laura Sue D'Annunzio, Challenges and Strategies of Matrix Organizations: Top-Level and Mid-Level Managers' Perspectives, Human Resource Planning (2005)
- Thomas Sy and Laura Sue D'Annunzio, Challenges and Strategies of Matrix Organizations: Top-Level and Mid-Level Managers' Perspectives, Human Resource Planning (full text) (2005)
- McKinsey & Company, Revisiting the matrix organization, McKinsey Quarterly (2016)
- Stanley M. Davis and Paul R. Lawrence, Problems of Matrix Organizations, Harvard Business Review (1978)
Questions
In Sy and D'Annunzio's study of 294 managers in seven multinationals, the top five were misaligned goals, unclear roles and responsibilities, ambiguous authority, the lack of a matrix guardian and silo-focused employees. Most reduce to decisions that were shared between two lines without being assigned to either, which is where the structure slows.
Because a decision with two legitimate claimants and no named owner has nowhere to go except upward or into a meeting. Each escalation adds days and loads the executive team with calls it should not be taking. Assigning the decisions that carry the plan to one line, with the other consulted, removes most of the delay.
They deliver what they are built for. McKinsey's analysis of Gallup data found matrixed employees reported better collaboration than non-matrixed peers, while being less clear about what was expected of them. A matrix works when the organisation puts the clarity back deliberately, through assigned decisions and a role that watches how the matrix performs.
A role, named in Sy and D'Annunzio's research, that owns the health of the matrix itself: it monitors how the structure performs, surfaces decisions that stall between the lines and settles ownership disputes that neither line will raise. In practice it is often held by a chief operating officer with a standing review of stalled and duplicated decisions.
Not before the decisions have been assigned. A matrix that slows has usually skipped that step, and a restructuring that skips it too will produce the same stall in a new shape. Assign the decisions that carry the plan, review what still escalates after two quarters, and only then ask whether the design itself is wrong.