Why does accountability keep breaking down in our matrix, when everyone in it is capable and trying? Some version of that question reaches me at almost every leadership conference and quarterly business review I speak at. Solving matrix organisation accountability problems is not, in my experience, a people problem. It is a design gap, and it is fixable without touching the org chart. This is a specific case of a wider pattern I cover at the cross-functional execution work I do with organisations, but a genuine matrix, with dual reporting lines, deserves its own answer.

If you are a chief operating officer, a divisional or functional leader, a chief transformation officer, a chief of staff, or you run an enterprise PMO trying to explain to your CEO or CHRO why the same intersection keeps producing drift, this is written for you. It is the kind of problem that surfaces at functional leader forums and program reset days, usually described as “we need better collaboration,” when what is actually missing is a decision.

A Matrix Creates Conflict by Design, and That Part Is Fine

Organisations do not build a matrix by accident. They build one because they genuinely need two things that pull in different directions: functional consistency, so the same discipline applies everywhere, and market or customer responsiveness, so a local team can move at the speed its situation demands. Both needs are real and legitimate. A matrix is an honest attempt to hold both at once, and most of the time it works, because most decisions do not force a choice between the two.

The design is not the failure. The failure is what most matrices leave unsaid: what happens when the two legitimate claims land on the same person, on the same day, in direct conflict. Almost no matrix I have looked at closely has an answer to that question written down anywhere. It is assumed the two leaders involved will sort it out, or that the person in the middle will use good judgement. Neither assumption holds up under real pressure, and the gap between the design and the missing answer is where accountability quietly dissolves.

The Conflict Ends Up with the Person Least Able to Resolve It

Here is a composite example I use often, because it is so ordinary it barely reads as a story. A regional operations manager reports into a functional line and a business line. The functional line tells her to standardise a process, because consistency across regions is what it is accountable for. The business line tells her to move fast for an important customer, because responsiveness in her region is what it is accountable for. Both instructions are correct, and both come from someone whose opinion of her affects her career.

She did not create this conflict and has no mechanism to resolve it. So she does the only rational thing available: she keeps both relationships intact, satisfies neither instruction fully, and the work drifts into a compromise nobody asked for. She is arbitrating a decision that was never hers to make, because the structure above her pushes the real decision too far down the organisation. Calling that a lack of accountability is unfair, and worse, it points everyone’s attention at the wrong layer of the problem.

Why Drifting Is the Rational Response to Two Legitimate Bosses

When two people who both control your future disagree and no one above them has said which one wins, drift is the safest available move. Picking a side burns a relationship you need. Escalating feels like admitting you cannot do your job. Blending the two instructions into a half measure looks, from where she is standing, like the mature choice, and it is also how a customer commitment slips or a standard quietly erodes.

This is the mechanism behind what I call Ownership Fog: the condition where every task in a cross-functional initiative has a named owner and the outcome itself has none. In a matrix, fog forms not because no one owns anything, but because the outcome depends on two owners agreeing, and nothing forces that agreement to happen on a specific day.

Decide Which Line Wins Before Tuesday Afternoon

The instinct here is usually to flatten the matrix or redesign the reporting lines. I would push back on both. Restructuring a matrix is expensive and slow, and in almost every case I have seen, it produces a differently shaped version of the same unresolved question, because the tension between consistency and responsiveness has not gone anywhere. You have just moved the argument to a new set of boxes.

The fix is not structural; it is a decision made in advance. For each recurring type of conflict, someone senior enough to own both sides needs to say, in writing, which line wins by default, and who arbitrates the exceptions. A regional operations network probably has four or five of these: pricing exceptions, staffing during peak demand, process deviations for an important customer, timeline trade-offs when a functional standard changes. Each gets a default and a named arbiter, and that is a short document, not a program.

Publish the Defaults, Name the Arbiter

The publishing matters as much as the deciding. A default that lives in one senior leader’s head is a preference, not a default, and the person at the intersection has to guess at it under pressure. Write the defaults down, put them where people can find them without asking, and revisit them on a schedule. I cover the underlying decision rights method in full, for any organisation with functions, in a companion piece on closing accountability gaps; this piece has a narrower job, which is what happens to that method when two reporting lines are involved instead of one.

A default that everyone knows beats a negotiation that everyone dreads. The operations manager in my earlier example is no longer picking a side between two people who evaluate her. She is applying a rule a more senior person already decided, in a calmer moment.

The 48 Hour Rule for Conflicting Instructions

Defaults handle the recurring cases. They will not cover everything, and should not try to. What handles the rest is a rule about time, not content, and it is the sharpest version of a principle I call The First 48 Hours: a decision is only believed when something visibly stops, starts or changes in the first 48 hours after it is made. Applied to conflicting instructions in a matrix, the test is not whether something eventually changed; it is whether the conflict got resolved before it aged into drift.

The rule is specific. When a person at the intersection receives two conflicting instructions, both leaders have 48 hours to agree a resolution. If they have not, it goes automatically, and without prejudice to either of them, to the person they both report to. No one has to ask for that escalation, because it triggers itself. This works for the same reason The First 48 Hours works everywhere else: it converts a vague commitment to “work it out between yourselves” into a visible, dated event with a clock attached.

The effect is upward, and it is the part leaders underestimate. Two senior people who know an unresolved conflict lands automatically on their shared boss’s desk on Thursday have a strong incentive to resolve it on Wednesday, on their own terms. The clock changes behaviour higher up the organisation than the person standing at the intersection ever could.

What to Fix Instead of the Org Chart

If your matrix keeps producing this pattern, the fix is not a redesign project. It is three things, done in sequence: identify the recurring conflict types your matrix actually produces, publish a default winner and a named arbiter for each, and install the 48 hour rule for everything that falls outside those defaults. None of that requires new boxes on a chart. It requires senior leaders to make decisions in advance that they were previously leaving for someone else to improvise under pressure.

I talk through this pattern, and the broader discipline of turning decided things into done things, in the keynote work I do on accountability and ownership, usually for leadership conferences and functional leader forums where this intersection keeps coming up in corridor conversations before the session starts.

If your matrix structure keeps pushing conflicts down to people who cannot resolve them, and you are preparing for a quarterly business review or a leadership forum where that pattern needs naming honestly, check my availability for that date. The case worth making in that room is simple: the intersection is costing you good people, and deciding in advance is cheaper than arbitrating one conflict at a time.