Matrix organisations are good at the things matrix organisations are good at. They are not good at deciding who actually gets to call it when two functions disagree. Which is a problem, because most consequential commercial decisions are exactly the decisions two functions disagree about.
The argument is usually about decision rights. The fix is usually about decision queues.
The standard mistake
The standard fix when two functions disagree is to redraw the RACI — the responsible / accountable / consulted / informed map. This produces, over a six-month exercise, a document that says who is accountable for what. The document is usually accurate. The document usually does not change behaviour.
It does not change behaviour because the original problem was not that nobody knew who was accountable. The problem was that, in practice, the decision was being made by whoever had the last meeting before it had to ship. Which was rarely the person the RACI named as accountable, and very often the person the RACI named as consulted.
The question that works
The question I’ve seen change things on more than one occasion is shorter than the RACI exercise. It is: who do we want this decision to be made by, when there is half an hour left on the clock and three of us still disagree?
Notice what the question does. It strips out the polite version (who should be accountable in principle) and replaces it with the operational version (who, in the last thirty minutes, gets the casting vote). It collapses the abstraction of accountability into the specifics of how a real disagreement, on a real clock, actually resolves.
The good answer is the answer that survives the half-hour-on-the-clock test.
About a quarter of the time, the answer is the person the RACI named. In which case the RACI was right, and the question costs nothing. About a quarter of the time, the answer is the person nobody had named, and the conversation that follows is the most useful conversation the team has had that quarter. The other half of the time, the answer is “we genuinely don’t know”, which is a finding worth its weight.
Why this beats the formal redraw
The half-hour-on-the-clock test works for two reasons. First, it forces the room to imagine an actual disagreement, which makes the answer concrete. Most decision-rights questions are answered in the abstract, where the answer is easy. They are then exercised under pressure, where the abstract answer breaks down. Asking about the pressure case in advance is cheap.
Second, it is a much faster exercise than redrawing the matrix. The matrix takes six months and produces a document. This conversation takes forty-five minutes and produces a decision. The matrix is a deliverable; the conversation is a decision.
One worked example
At a previous business, the disagreement was between commercial and supply chain on the depth of a promotional plan. Commercial wanted aggressive depth on a watchlist; supply chain was worried about the service implications. The RACI said commercial was accountable. The actual decision pattern, when we looked at twelve quarters of promotional plans, was that supply chain’s concerns were dropping the depth of about a third of the watchlist by the time it went live.
We didn’t redraw the RACI. We had the conversation: with half an hour on the clock, who calls this? The answer was that commercial called the depth, supply chain called the timing window, and disagreements about depth-given-timing went to the commercial director. That was the operational rule we’d been pretending to have for a year. Writing it down made it actually true.
Not every disagreement needs a process. Most need a name. The half-hour-on-the-clock test gives you the name.