The meeting has happened before. Two people who both know the operation could sit across from each other, each with a model, a spreadsheet, or a plan that says something different from the other's. Neither is bluffing. Neither is wrong in any way the other can point to. The meeting ends the way it started, and the decision sits there while the cost of waiting compounds, because nobody in the room has a way to settle an argument between two people who are each, by every measure that matters, right.
The easy read is that this is a people problem. Two strong-willed, experienced people each think they're right, so the room treats it as a clash of judgment, or seniority, or whoever's louder in the meeting, and tries to resolve it with more argument or a decision from whoever outranks the other. It rarely works. The disagreement comes back the next quarter, just with new numbers behind it, because argument doesn't reach the reason two accurate people can disagree.
The real cause is more structural than anyone in the room wants to admit. In a tightly coupled operation, each person is reasoning from a real, accurate view of their own piece of it, and two accurate views can point in opposite directions. The head of one function sees the true state of their area. The head of another sees the true state of theirs. Neither is wrong about what they know. Neither one knows the whole. When the parts of the operation interact the way they do in any complex system, the locally right answer in one area can be the wrong answer for the system as a whole, and there's no way to see that from inside either function alone. Researchers who studied operators of interdependent infrastructure systems—power, steam, water, and IT—found the same pattern under conditions of high uncertainty: conflict between operators wasn't a communication failure; it was what tightly coupled systems produce when nobody shares a view of how the pieces connect.
A second layer makes the disagreement harder to break. Both people aren't just partial; they're confident, and on hard problems confidence tends to run ahead of accuracy. Research on overconfidence has found this repeatedly: on difficult tasks, people consistently overestimate how well they've actually got it right. Neither expert will yield, because each trusts a partial view more than the evidence supports, and each experiences that view as obviously correct. The room can't settle that with a more confident assertion from either side. It never had a tie-breaker to begin with.
The room can't move from competing assertions to a choice it can stand behind by debating harder or deferring to whoever has the higher title. What it needs is something neither person currently has: a single picture of the operation accurate enough that both sides can test their own assumptions against it. Not a tie-breaker vote. Not a louder argument. A shared, testable representation of the thing they're actually disagreeing about.
This picture has to do more than sit still and look neutral. It has to hold both partial views at once, show how they actually interact when the operation runs, and let each person see what happens to the whole system when their preferred option plays out. Once the disagreement can be run instead of argued, it stops being a standoff between two confident people and becomes a comparison between two testable options. The conversation changes shape entirely. It's no longer about who's right. It's about which option performs better against the operation both people already agree they're looking at.
The shift from arguing to testing is also what makes the decision hold. A choice reached this way carries its own reasoning, so it doesn't quietly reopen next quarter once the details have faded from memory. When someone challenges it later, they're testing that reasoning against the picture, not relitigating a debate nobody wrote down.
The disagreement was never really a judgment problem. Two capable people had two accurate, partial views, and no shared picture either one could test their assumptions against, which is what a judgment problem gets mistaken for when nobody can name what's actually missing. Give the room that picture, and something changes fast: people who were certain of opposite things start reasoning from the same operation, and convergence stops requiring anyone to back down.
Next time a decision stalls on disagreement, the useful question isn't who's more senior or whose numbers sound more confident. It's whether anyone in the room can point to a single tested picture of the operation that both sides already accept.
A large apparel company faced the same standoff. Stakeholders needed to be convinced that a set of capacity requirements was the right call, and no spreadsheet alone would do that. SimWell built one model both sides could test scenarios against, which is what let a divided group commit.