How to get people to disagree with you

Cerno is a private AI boardroom where disagreement is structural — a Devil's Advocate whose entire role is to attack the leading option, and dissent recorded in the brief rather than smoothed away. This guide is about the same problem with human beings, which is where most owners actually meet it.

Why nobody disagrees with you

It's rarely fear, and it's almost never that they agree. Four ordinary mechanisms do most of the work.

Agreement is free; disagreement has a price. Saying "sounds good" costs nothing. Saying "I think that's wrong" costs a small amount of social capital, risks being wrong in front of someone, and takes longer. Given an easy option and a costly one, most people take the easy one most of the time — including people who genuinely would speak up about something serious.

You signal the answer before they speak. "I think we should do X — what does everyone think?" has already told the room what you want. So has your tone, the fact that you've clearly been thinking about it for a week, and the slide you've made. People aren't responding to the question; they're responding to the answer you've shown them.

They're paid, or they're fond of you. Your accountant, your solicitor, your contractor and your friends all have relationships with you that a hard challenge could strain. Not corruptly — they just have a mild, unexamined interest in the conversation going pleasantly, and so do you.

Nobody's job is to object. In most small businesses, challenging a decision isn't anyone's role. It's an act of individual initiative, and initiative is uneven. Where disagreement is somebody's assigned task — an auditor, an opposing counsel, a Devil's Advocate — it appears reliably, from the same people who'd otherwise have said nothing.

Six things that work

1. Ask for the case against, not for an opinion

"What do you think?" gets agreement. "Give me the three strongest reasons this fails" gets reasons — because you've assigned a task rather than requested a judgement, and completing it is being helpful rather than being difficult.

The number matters. Ask for three and you'll get three, including the one they wouldn't have volunteered. Ask for "any concerns" and you'll get none, because zero is a valid answer to that question.

2. Don't say what you think first

If you have a view, hold it until everyone else has spoken. This is the single highest-return change and the hardest to do, because you called the meeting precisely because you have a view.

Where you can, get positions in writing before the conversation: two lines by email, sent to you rather than to the group. That removes both your anchor and the first speaker's. It's the same logic as the independent-writing step in a pre-mortem, and it works for the same reason.

3. Assign the objection

Name someone and give them the job: "Sam, for this conversation you're arguing against. Your job is to make the strongest case that we shouldn't." This removes the entire social cost at a stroke — nobody's being difficult, they're doing what they were asked.

Rotate it, so it doesn't become one person's identity. And thank them visibly when they do it well, because the rest of the room is watching to see what happens to the person who objected.

4. Ask what would have to be true

Instead of "do you agree?", ask: "what would have to be true for this to be the right call?" Then: "how confident are we in each of those?"

This reframes disagreement as analysis. Nobody has to say you're wrong; they just have to say a thing you're relying on is shaky. People who would never challenge a decision will happily point out that an assumption is untested, and the effect is identical.

5. Make it cheap to be wrong

If disagreeing is only safe when you turn out to be right, people will only disagree when they're certain — which is exactly when you least need them to. Say plainly that you want the objection even if it turns out to be nothing, and then behave that way when it is nothing.

The real test is what happens the first time someone objects and is wrong. If it's awkward, or it comes up again later, you've taught everyone watching not to bother.

6. Go outside the circle

The people closest to the decision are the ones with the most invested in it. Someone one step removed — a peer in another business, a supplier who's seen a dozen firms like yours, your accountant asked a direct question — has no stake in the outcome and no relationship to protect in the moment.

Ask them something specific. "What's the most likely way this goes wrong?" gets a real answer. "What do you think of my plan?" gets encouragement.

What doesn't work

"Feel free to push back." An invitation to disagree is not a mechanism. It costs you nothing to say and changes nothing about the social cost of taking it up. Every boss who has ever been disagreed with least has said this.

Asking in a group, cold. The first person to speak sets the direction, and in a small business the first person to speak is usually the most senior after you. Get the independent positions first, then talk.

Asking after you've decided. People read the room accurately. If the decision is visibly made — you've told a customer, you've ordered the thing, you've announced it — objections are now criticism of you rather than input on a choice, and they will not come.

Rewarding the objection you agree with. If challenges that turn out to be right get praised and ones that turn out to be wrong get remembered, you've made disagreement a bet. People stop betting.

Where a structured board fits

Every technique above depends on someone being willing, available and honest at the moment you need them — which is precisely the thing a small business can't rely on at eleven at night. In Cerno, the disagreement is structural: advisors argue from different lenses, a Devil's Advocate takes the case against as its assigned role, and the dissent is written into the brief rather than depending on whether anyone felt like raising it. The techniques here are what it's automating; use both.