You can win an argument and make the other person less willing to listen to you next time. That’s an expensive victory. You’ve improved your position in the debate while making the relationship less useful for discovering anything together.
When someone asks how to change another person’s mind, I want to know what change would actually look like. Do you need agreement about a fact? A different decision? A willingness to run a test? Or do you want them to admit that you were right all along?
The last objective often interferes with the others. If a conversation becomes a public ceremony of surrender, the other person has more to protect than the original belief. A better approach gives them room to examine an idea without making their dignity the price of admission.
“They’re against innovation” is a poor starting point. It replaces a specific concern with a convenient identity. Imagine a manager who opposes a new software purchase. They might believe the tool is unreliable, that training will consume too much time, or that the team hasn’t used the current system properly.
Those are different propositions. Evidence that the software works won’t answer a concern about training time. A demonstration of features won’t resolve distrust of the purchasing process. If you don’t know which claim matters, you can produce excellent arguments for a conversation nobody else is having.
Ask them to explain the position in their own words. Then summarize it without caricature: “You think the new system could be useful, but the team can’t absorb another transition this quarter.” Check whether that’s fair. If it isn’t, ask what you’ve missed.
Don’t use this exercise as a hidden delay before delivering your prepared speech. You should be willing to discover that the objection is better than you expected. Changing minds is a two-way possibility. If your method only permits the other person to revise, you’re organizing a conversion attempt rather than a joint inquiry.
People can hold an opinion for several reasons at once. It may describe what they think is true, protect them against a feared outcome, or express loyalty to a group. You don’t need to speculate about their unconscious motives. Ask what they see at stake.
Useful questions include: “What would worry you most about being wrong here?” and “What experience led you to that conclusion?” These questions work best when you want the answer. Asked with a smirk, they become a more elaborate form of dismissal.
In our hypothetical software conversation, the manager might describe a previous transition that disrupted customer service. Now you know why another feature demonstration isn’t enough. The proposal needs a plan for protecting service during the change. You can address that concern without claiming the earlier decision was foolish.
Sometimes a belief has a practical job that should survive even if the belief changes. A policy might be based on an outdated assumption while still protecting against a real risk. Keep the protection and revise the mechanism. People have more room to reconsider when you aren’t asking them to abandon a legitimate responsibility.
“How can we improve response time without disrupting service?” is easier to work on together than “Why won’t you approve my software?” The first question includes both the desired improvement and the feared cost. It creates a problem neither person can solve merely by repeating their position.
Shared questions must be honest. You can’t promise collaboration while treating only your preferred answer as acceptable. If there are nonnegotiable constraints, name them. A team can explore delivery options within a fixed budget, but it shouldn’t pretend the budget is open when it isn’t.
Agree on how to evaluate possible answers. What evidence would show an improvement? What downside would make the proposal unacceptable? Who needs to be involved in the decision? Defining these conditions early reduces the temptation to move the goalposts later.
This connects to checking the story you want to believe. A fair test should be capable of disappointing you. If every possible result proves your idea was correct, the exercise is arranged to protect a conclusion rather than investigate it.
Facts matter. They also arrive inside a conversation between people who interpret relevance differently. A carefully chosen question about experience can help reveal that difference before you argue about numbers.
Field experiments by Joshua Kalla and David Broockman tested conversations addressing exclusionary attitudes. In that setting, conversations that included a nonjudgmental exchange of personal narratives produced reductions in those attitudes, while arguments alone did not in one experiment. The effects were modest and the context specific. This is evidence for examining how a conversation is conducted, not a universal recipe for changing any belief.
My practical lesson is to let people explain how they arrived where they are. Then offer relevant evidence or a truthful example that helps examine the question. A story can clarify a consequence that an abstract claim obscures. It shouldn’t be used to make one vivid case stand in for the entire pattern.
Keep the distinction between understanding an experience and establishing a fact. Someone’s bad encounter with one provider explains their concern; it doesn’t prove every provider will fail. You can respect the experience and still investigate whether the conclusion travels as far as they think it does.
Suppose the manager agrees that a limited test would be reasonable. This is a terrible moment to announce, “So you finally admit I was right.” You’ve just made future agreement more socially expensive.
Instead, describe what changed: “We now have a way to test the tool without moving the whole department.” The revised decision follows new conditions. No one has to pretend that their earlier concerns never existed. The conversation can move forward without rewriting history.
Often the useful change is smaller than a complete reversal. Someone moves from certainty to curiosity, from refusal to a pilot, or from an accusation to a request for records. Treat that change accurately. Don’t inflate it into agreement they didn’t give.
Privacy can matter, too. If you’re correcting a person’s claim in front of people whose respect they value, the audience becomes part of the problem. Some errors need public correction because others are relying on them. Even then, correct the claim precisely and avoid using the moment to perform superiority.
Afterward, a private conversation may allow more exploration. The aim is to reduce unnecessary humiliation while protecting shared accuracy. Neither goal requires you to pretend false information is acceptable.
Some disagreements persist because both people understand the evidence and value different outcomes. One person wants maximum growth; another wants a predictable workload. More facts about the market won’t automatically settle how much disruption either person is willing to accept.
Name the difference. “It sounds like we agree that this could increase revenue, but we disagree about whether that gain justifies the operational risk.” Now you can discuss tradeoffs instead of accusing each other of failing to understand.
Ask whether there is a design that serves both priorities sufficiently. A smaller launch, a later date, or a narrower customer group might work. If there isn’t, the decision may require a legitimate authority, a negotiated agreement, or a respectful refusal to proceed.
Not every conflict should be solved by persuasion. You may need a boundary when someone repeatedly ignores an agreement. You may need a formal process when responsibilities are disputed. The skill of resolving conflict includes recognizing when another argument won’t solve the structural problem.
Pay attention to the pace of the request. An opinion that developed over years may deserve more than one lunch conversation. You can make one useful distinction, share one relevant source, and let the person examine it privately. Repeatedly asking whether they’ve changed their mind can turn an invitation to think into a demand to report compliance.
There is also a difference between making evidence available and burying someone under it. If they question a specific cost assumption, send the calculation that addresses that assumption. Sending thirty unrelated links gives them a larger task without necessarily giving them a clearer reason to reconsider. Choose the smallest piece of evidence that can honestly carry the claim you’re making.
Ask permission before scheduling another round. “Would it be useful to revisit this after you’ve checked the figures?” respects both attention and ownership. If the decision affects your responsibilities, explain why a follow-up is needed. You can be clear about a required decision without pretending that ongoing personal debate is mandatory.
Before ending the conversation, summarize the current position. What do you agree about? What remains unresolved? What information or action would help? Who will do what, and when will you revisit the issue?
A useful ending might be: “We haven’t agreed to buy the system. We’ve agreed to test it with one team and review service quality alongside time saved.” That’s specific enough to prevent each person from leaving with a flattering but incompatible account.
If neither person is willing to name evidence that could matter, pause. Continuing may only deepen the pattern of attack and defense. You can say, “I don’t think we’re learning anything new right now. Let’s stop and return when we can discuss the actual decision.”
Also keep track of your own behavior. Did you ask a real question? Did you recognize a valid concern? Did you update anything you believed? A conversation aimed at understanding people should include the person doing the persuading.
Changing someone’s mind is always partly outside your control. What you can control is whether you create a fair opportunity for reconsideration. Make the claim clear, understand the stakes, and offer evidence that bears on the decision. Let the other person do their own thinking.