Disagreement Is Not a Request for Another Notebook

Someone objects, and the technically trained reflex is “let me look into that and come back,” followed by two weeks of evidence. This works when the disagreement is genuinely about a fact. It usually is not, and the cost falls in one direction. While the analysis is being extended, the incumbent keeps running as the room’s collective fallback. A request for more analysis that cannot change the ranking is not neutral. It is how the rule already in place survives a challenge it might lose on the merits.

Name the fight

Before doing anything, establish what kind of disagreement this is. There are six, and only some of them yield to evidence.

Factual. The parties do not agree about what happened. Which items stopped vehicles last year is a factual question, and it calls for a query rather than a platform.

Model. The parties do not agree that the mechanism in the write-up is the mechanism in the world. “The class-based rule will be bypassed by lunch” is a model of behavior: test it, or write it as a condition that reverses the ranking.

Assumption. The parties disagree about an input near a flip point. Show whether the ranking moves. If it does not move within any value either party would defend, the disagreement is about something else.

Objective. The parties do not agree what better means. A better forecast will not help, and the work moves back to economics and to interests.

Incentive. The parties agree privately and cannot say so, because the pick damages a scorecard. Further analysis serves mainly to legitimize the delay.

Risk. The parties do not agree how bad the bad outcome is, or who bears it. Show the tail rather than averaging it away, because a point estimate is where this kind hides.

When the kind is not obvious, one question usually settles it: if I could produce any number you wanted this afternoon, would you change your mind? If the answer is no, further modeling will not help. A related signal: when each new cut of the data confirms both sides in what they already believed, they are not disagreeing about the quantity being estimated.

State the classification as a complete thought: “we disagree about whether bypass can be closed, and if it cannot, the ranking changes.” Then ask the parties to point to the part of the write-up they dispute: the choice, a reason, a condition that would reverse it, or the rule. If they cannot point to any of these, the disagreement is not technical.

Agree the criteria before you argue the answer

A disagreement that will not converge is often one in which the criteria were never settled. An objection on cash is answered and moves to implementation risk, then to new parts, and the conversation cannot be won because nothing counts as a sufficient answer.

Agreement on criteria is nearly free before anyone knows which alternative they favor, and nearly impossible afterward, when a proposal to rank on cash and stranded jobs is heard as a proposal to win. So circulate the criteria a week early, on their own, framed as a question: the replenishment rules will be ranked on cash released and stranded jobs over twelve months, planner workload matters but will not decide it, and anything Parts Planning cannot run is out of scope. What should this turn on, and is anything missing?

In the meeting, an objection is then either evidence about an agreed criterion or a new criterion, which is legitimate and can now be named as such.

Give the disagreement to its owner

When the parties will not converge on criteria, stop producing evidence and write the disagreement down precisely enough that the person who owns the tradeoff can decide it. The note has four elements: the range the analysis supports; each party’s criterion and where the range falls against it; a statement that further analysis would not move the range across either criterion; and the name of the person whose decision it therefore is. Send it to the parties first and ask each to correct any misstatement of their position. Nobody is surprised, and the note becomes a shared description of the disagreement rather than your verdict on it.

The question that justifies stopping is: what result would change your position? When nobody can answer it, declining to continue does take a side, against an indefinite delay whose effect is to keep the incumbent running. That is a side you are entitled to take.

Get through, do not score

A ranking that is right and unheard has the same effect as one that is wrong. When somebody is hot, because of a stranded vehicle, a scorecard, or a number they promised publicly, more reasons spoken faster sound like an attack. Stop presenting. Restate what they are protecting until they recognize it. “You are worried that if we touch the pumps you eat the weekend, and the ranking on the page does not yet show you that you will not.” Then wait. If they say yes, they can now hear the rest. If they say no, you have just found the real interest.

Appeal to the work people take pride in: a class the planner helped specify is a rule the planner can defend, and one presented as a correction of the planner’s judgment will be padded before it fires. Put the commitment in the form that is cheap to accept: “two centers, ninety days, and a review on 3 July with the power to stop it” is a challenge a proud operator can take.

Scoring the point feels like analysis. It is how you lose the next ten minutes and, often, the freeze.

Let people change their minds cheaply

Keep the disagreement on the recommendation. A director who argued for the blanket cut in March is defending the person who argued for it as much as the cut, and needs a way to withdraw with dignity. Attribute the earlier position to the information available at the time, which is nearly always accurate: on what was known in March, before anyone had the expedite records, the blanket cut was a reasonable reading. Credit by name the objection that improved the note, and never return to a concession at the next meeting as evidence of having been right.

Then put your weakest reason on the page, in your own words, before anyone else finds it. Once the write-up says the classes will need rework when a new vehicle platform launches, a colleague who raises it is agreeing rather than defeating you.

Neither hide uncertainty nor perform it

Drowning the pick in caveats is the form hiding takes among careful people, and it gives a hostile room its exit. Bring the flip points, what each party is protecting, and the rule, and ask which part they think is unfinished: the choice, the evidence, or the rule. Equally, do not bury a risk that would kill the pick you prefer.

The objection you cannot answer

Bluffing works until the objector checks. Collapsing, offering to study it while the decision slides past the freeze, feels humble and is the most expensive option. Instead, say four things aloud: that the answer is not known; what kind of question it is; what would settle it, and roughly at what cost; and whether the answer could change the recommendation. The fourth is the one people forget. If a regional manager asks whether the concentration holds in Canada, the honest answer may be that a query would settle it by Thursday, and that unless Canada differs sharply it would not reassign the items that drive the ranking, so commit now and check Thursday. If it could change the recommendation, say so: a live unknown with a named test and a date is still a recommendation.

When to concede, and three concessions to refuse

Sometimes the objection is simply good: dealer operations shows that missed-job cost is more uniform across items than believed. Redo the ranking rather than defend last week’s.

Three concessions look gracious and are not. Do not accept two incompatible positions as constraints in order to end the meeting. Do not replace a rule with an undertaking to monitor it unless the monitoring has an owner and a trigger. And do not agree to a method that cannot change the ranking because it would make the objector feel the work is rigorous.

When they will not play

Some rooms run a different game: a deadline that is only a meeting, a take-it-or-leave-it on a number that is still a position, a friendly visit afterward to strike the reversing condition in private. Do not match the tactic. Restate it as an interest or a fallback and put it on the page; a condition struck in the hallway is a second decision, and it gets written down as one.

After the meeting

Record the disagreement by its kind, not as “alignment pending”: who runs the query and by when, the new flip number, or what will not be modeled. Update the note the same day; a note that still carries the pre-meeting recommendation after a real concession is a second decision in circulation. If the meeting ended in a request for more analysis that cannot change the ranking in time to matter, name it as a delay and record the incumbent’s continued operation as its cost.

A five-minute test

Take the last objection you answered with more analysis. Classify it as one of the six kinds. Then write the answer its author would have given to “if I could produce any number you wanted this afternoon, would you change your mind?” If the honest answer is no, the next deliverable is a one-page note to the owner, not another notebook.