Your Unpopular Decision Isn't Done When You Announce It

The head of engineering makes an unpopular decision on a Thursday. Starting next sprint, an AI agent reviews every pull request first. Humans review after the agent, and nobody leaves style or lint comments by hand anymore. Two senior engineers argued against it in the meeting. They lost. On Friday the decision goes out in Slack, and by Monday both of them have given the post a thumbs-up.

Three weeks later, cycle time is worse than before the change. The two seniors are doing full line-by-line reviews alongside the agent. Every few days one of them posts “the agent missed this” in the team channel with a screenshot. Nobody refused the decision, and it is in effect. The team is paying for code review twice.

Most leaders read that as resistance and start thinking about how hard to push. I think the leader stopped working too early. The announcement felt like the end of the job, and it was the halfway point. After an unpopular decision, the leader still owes work on how it lands, and that work can happen without reopening the decision at all.

The Assumption: Once You Decide, Your Job Is to Hold the Line

The phrase most teams use for this is “disagree and commit,” which Jeff Bezos made famous in his 2016 letter to Amazon shareholders. The idea is sound: argue hard while the question is open, and once the call is made, everyone works to make it succeed, including the people who argued against it.

In practice most leaders run the first half and hand off the second. The disagreement happens in the meeting. The commit is treated as the other person’s job. If they are still unhappy a week later, that is a maturity problem, and the leader’s job is to hold firm until they get over it.

The reason is simple. Asking someone how they feel about a decision you already made looks like reopening it. Leaders worry that one sympathetic question turns into a negotiation, the negotiation turns into an exception, and six weeks later the decision exists only on paper. So they announce, hold, and wait.

Part of that instinct is right. The bar should not move because people are upset. A decision that bends for whoever complains loudest teaches the whole team that complaining is how you get your way.

The Crack: The Decision Held and the Work Slowed Down

Go back to the pull requests. Nobody broke a rule. Both seniors said they were committed, and they did what the decision required. The agent runs on every PR. They also kept doing what they did before, which doubles the review time and leaves a trail of screenshots that proves the agent is imperfect.

The agent is imperfect. So are human reviewers, and the team made the call knowing both, so the screenshots add nothing the meeting did not already weigh. A commit made only in words tends to look like this three weeks in.

The signs are easy to spot once you look. People do the old work in parallel with the new work. They post “told you so” with receipts in a public channel. They save their objections for the retro, where they can be raised in front of everyone, instead of the one-on-one. Someone is silent in the meeting and then sends a long message to a peer afterward.

A leader who sees this usually concludes the people are fighting the decision. Some are. More often, I think, they are carrying the cost of a call nobody is helping them carry. The leader made the decision on Thursday and went back to other work on Friday, and the two people who lost the argument woke up Monday holding all of it.

The Investigation: What the People Who Lost the Argument Need

When you sit down with the person who lost and ask what is going on, in my experience they rarely ask for the decision to be reversed. They usually know it is not coming back. What comes up instead tends to fall into three needs.

The first is someone saying out loud that the cost lands on them. Those two senior engineers spent years getting good at code review. It was part of how the team saw them and part of how they saw themselves. The decision changes what that skill is worth on this team, and nobody mentioned it. When the leader says nothing about that loss, the two of them conclude they are the only people who noticed.

The second is a say in how, even after losing on what. The decision settled whether the agent reviews first. It left open which repos go first, how the team handles the agent’s comments when it is wrong, who tunes the instructions it reviews against, and a dozen smaller questions like those. The people with the most review experience are the best people to answer them.

The third is some condition under which the call gets looked at again. A written trigger, such as “if escaped defects go up over the next six weeks, we revisit,” turns a decision that feels like a decree into a bet the team is making together.

William Bridges spent a career on this gap. In Managing Transitions, he separates the change, which is the event you announce, from the transition, which is what people go through internally afterward. Transition starts with an ending: the thing people are losing. Leaders announce the change and skip the ending, and then they are surprised that people keep acting like the old way still exists.

Aristotle gave the goal a name. In Book IX of the Nicomachean Ethics, he describes homonoia, usually translated as concord. He is careful to separate it from simple agreement of opinion. People can agree in opinion about anything, he says, even the stars, and strangers can do it. Concord is about practical matters. People are in concord when they choose the same course and carry out together what they decided. He calls it political friendship, the bond that holds a city together, and he says it exists among decent people. One side cannot produce it alone.

Then he says when it breaks. Concord does not last among people who want more than their share of the benefits and less than their share of the work. Each one watches the others to keep them from shirking, and the common work falls apart.

Read that from the leader’s chair. The leader made the call and got the benefit: a faster pipeline, cleaner reviews, and credit for moving the team forward on AI. Then the leader left the hard part of the transition to two engineers and went back to the calendar. By Aristotle’s measure, the person shirking their share of the work in that story is the leader.

The Revelation: Buy-In Gets Built After the Call, and Part of It Is Yours

Buy-in after an unpopular decision does not mean the people who argued against it now think you were right. Their opinion may never move, and you do not need it to. You need them to choose the same course and help carry it out. That is concord, and it is a much more reachable goal than agreement.

Before you try, make sure you can say in two plain sentences why the decision serves the team’s work. If you cannot, the problem sits upstream of buy-in, and no amount of follow-up will fix it.

If you can, getting there takes one question, asked with a clear boundary around it: “What can I do to get you more comfortable with this?” Comfortable here means able to do good work under the decision, whether or not they are happy about it or agree with it.

The boundary keeps the question from turning into a negotiation. The decision stays closed while everything around it is open, and you say both in the same breath so nobody walks out thinking the call is back on the table.

The leader has a lot to offer: the timing and order of the rollout, training and time to learn the agent’s habits, ownership of the review instructions, a written revisit condition with a date, and a standing check-in. A good share of the discomfort goes away once people have real control over the parts of a decision that are still open.

The one thing off the table is an exception to the decision itself. If the two seniors get to skip the agent because they were unhappy, the decision is over, and the rest of the team learns that the bar moves for whoever pushes hardest. You pay for comfort with your time and attention, and you keep the bar where it is.

The right move sits between two failures, leaving the people who lost alone with the decision and bending the decision to make them feel better: shared work on the open parts of a closed decision.

Shared work runs both ways. The people who lost the argument said they were in, and running the old process next to the new one is a choice they are making. Your help makes an honest commit easier, and the commit is still theirs to make. If you have done your share and the parallel work keeps going, it becomes a direct conversation about the job, and you can have that conversation with a clear conscience. The same holds when you are the one who lost: a manager who never followed up is a reason to go ask for the conversation yourself, and no reason to keep the old way running.

What you control is the offer and your share of the work. Whether they take it is up to them. Judge yourself by whether you did your part, because their mood on a given Tuesday is not yours to set.

Keep discomfort and new information apart. If someone comes back with a fact you did not have when you decided, such as a class of security bug the agent consistently misses in your codebase, that is a reason to look at the decision again. Treat it that way and thank them for it. If they come back with the same argument they made in the meeting, that is discomfort, and you help them carry it. Say out loud which one you think you are dealing with. People notice when a leader waves off a real fact as more complaining, and they notice when a leader caves to complaining and calls it new information.

The Application: Five Moves for the Week After an Unpopular Decision

1. Name what is closed and what is open in the same sentence. “The agent reviews every PR first. That part is decided. How we roll it out, which repos go first, and who owns the review instructions are all still open, and I want your help on them.” One message, both halves, so nobody has to guess.

2. Ask the question one-on-one. The announcement thread is the worst place for it, because people dig in harder on a position they took in front of others. Alone, they can tell you what bothers them. Something like: “I know you argued against this, and I know it changes how your review work counts on this team. The call stands. What can I do to get you more comfortable with it?” Then stop talking and listen to the whole answer.

3. Pay for comfort with your time, never with exceptions. Offer the training, the ownership, the sequencing, and the check-in. Hold the decision itself exactly where it is. If you notice yourself drafting an exception for one person, stop and ask what you are trying to buy with it.

4. Write the revisit trigger down, with a date. “If escaped defects rise over the next six weeks, we revisit. We look at the numbers together on November 25.” A written trigger turns a decree into a bet with an end date, and it gives the person who disagreed a legitimate way to be proven right. Most of the parallel work stops once people know there is a real way to reopen it.

5. Go back in two weeks and ask again. Most leaders ask once, if they ask at all. The second conversation is where you find out whether anything you offered helped. It also shows the person the first one was not a box you checked.

The step before this one is hearing people before you seal the decision, and some calls should never have been put to the room at all (Not Every Decision Is a Discussion). The five moves above apply after either one.

Final Thoughts

An unpopular decision has two parts. Making the call is the part leaders are trained for, and it usually takes an afternoon. Helping the people who lost the argument carry it takes weeks, and most leaders skip it because from the outside it can look like going soft.

The bar stays where you set it. The difference is that you stop leaving the cost of your decision with the people who argued against it. Aristotle’s concord asks for shared work on a course not everyone chose. The person who chose it owes the biggest share, and the people who argued against it still owe theirs. Close the decision, open everything around it, ask what would make it easier to carry, and keep asking.

If you want to become the kind of leader whose hard calls get carried instead of resisted, MasteryLab is built for that.

FAQ

How do you get buy-in after an unpopular decision?

Keep the decision closed and open everything around it. Meet one-on-one with the people who argued against it and ask what you can do to get them more comfortable. Offer what you control, such as timing, sequencing, training, support, and a written condition for revisiting the call. Do not grant exceptions to the decision itself. Then come back in two weeks and ask again.

Why does disagree and commit fail so often?

Most leaders treat the commit as the other person’s job. The disagreement happens in the meeting, the announcement goes out, and the people who lost the argument carry the cost alone. They say they are committed and then follow the letter of the decision while doing the old work in parallel. The commit holds when the leader keeps working on how the decision lands after it is made.

What is homonoia?

homonoia is Aristotle’s word for concord. In the Nicomachean Ethics he separates it from simple agreement of opinion. People are in concord about practical matters when they choose the same course and carry out what they decided together. He calls it political friendship and says it breaks down when people want more than their share of the benefits and less than their share of the work.

When should a leader reverse an unpopular decision?

When someone brings a fact you did not have when you decided, or when the revisit condition you wrote down trips. Discomfort calls for helping people carry the decision, while new information calls for looking at the decision again. Treat the two differently, and say out loud which one you are dealing with.