Playing Devil's Advocate: How to Use It Well (Without Being Annoying)

The decision took eleven minutes. Rosa's marketing team agreed to move the whole spring budget into one big influencer campaign. Everyone nodded. The deck looked great, the founder was excited, and the Friday afternoon meeting was running long.

Two months later the campaign had produced a lot of likes and almost no sales. In the retrospective, three different people admitted they had doubts at the time. One thought the audience was wrong for the product. One worried about putting everything into a single channel. One had quietly noticed that the influencer's past sponsored posts never mentioned prices.

Nobody said any of it out loud. Nobody wanted to be the person who kills the mood on a Friday.

That is the gap a devil's advocate fills. Done well, it is cheap insurance: a few minutes of deliberate pushback before a decision instead of months of regret after it. Done badly, it is the most annoying person in every meeting. This guide shows the difference, with a question bank, word-for-word scripts, and a worked example.

What "devil's advocate" means

To play devil's advocate is to argue against a position on purpose, usually one you may actually agree with, to test whether it holds up. The point is not to win. The point is to find the weak spots before reality does.

You will hear it in everyday phrases like "let me play devil's advocate for a second" or "just to be the devil's advocate here." Both signal the same thing: what comes next is a test of the idea, not a personal attack on it.

Where the term comes from

The phrase has a literal origin. In the Catholic Church's process for declaring someone a saint, an official was tasked with arguing against the candidate: questioning the evidence of miracles, raising doubts about the person's life and character, and making sure the case was not accepted just because everyone wanted it to be. His formal title was Promoter of the Faith, but he was popularly called the advocatus diaboli, the devil's advocate.

The office was formalized in the late sixteenth century, and its role was greatly reduced in 1983 under Pope John Paul II. The idea outlived the job: a group that badly wants an outcome needs someone whose task is to ask, "But what if we are wrong?"

Why playing devil's advocate works

Good people make bad group decisions all the time. Not because they are careless, but because a few predictable forces push groups toward agreement.

Groupthink

Psychologist Irving Janis coined the term groupthink in the early 1970s to describe how cohesive groups can reach poor decisions because members value harmony over critical evaluation. The warning signs are familiar: an illusion that the group cannot fail, pressure on doubters to fall in line, and self-censorship, where people keep their concerns to themselves. That last one is exactly what happened in Rosa's meeting.

A devil's advocate breaks the self-censorship. When one person is officially allowed to object, objecting stops being a social risk, and often two more people say "actually, I was wondering about that too."

Confirmation bias

We all tend to notice evidence that supports what we already believe and skim past the rest. If you want the new job to be a good idea, your brain helpfully collects reasons why it is. Arguing the other side forces you to look for the evidence you would naturally skip.

The premortem

Psychologist Gary Klein popularized a technique called the premortem. Before a project starts, the team imagines it is a year later and the project has failed. Then everyone writes down why it failed.

The premortem works for the same reason devil's advocacy works: it gives people permission to voice doubts, and it changes the question from "is this a good plan?" (where everyone defends it) to "how did this plan fail?" (where everyone gets creative about risks). You can think of a premortem as a whole team playing devil's advocate at once, and it is one of the best tools in this guide.

When devil's advocacy backfires

The technique has a bad reputation in some offices, and it is not undeserved. Here is how it goes wrong.

The contrarian for sport

Some people play devil's advocate in every meeting, on every topic. It stops being a test and becomes a personality, so the one time they spot a real problem, nobody listens.

The person who is "just asking questions"

Sometimes "I'm just playing devil's advocate" is a shield. The speaker gets to say something provocative or hurtful while disowning it in the same breath. If the "devil's advocate" position is really about another person's worth or identity rather than a plan, it is not a stress test. It is a way to say something without being accountable for it.

Assigned dissent is weaker than real dissent

Psychologist Charlan Nemeth spent years studying dissent in groups. One of her best-known findings is that authentic dissent, from someone who actually disagrees, tends to stimulate better and more original thinking than a devil's advocate who is merely assigned the role. When people know the objection is role-play, they can dismiss it easily. They may even become more confident in the original plan, because they "already considered the other side."

This does not mean you should drop the technique. It means you should use it to surface real doubts rather than replace them with a performance: focus on concrete assumptions, and actively invite the people who genuinely disagree to speak. Also watch for timing. Objections raised after the contract is signed are not devil's advocacy. They are complaining.

How to play devil's advocate well

Here is the method, step by step. It works in meetings, in one-on-one conversations, and inside your own head.

1. Announce the role

Say what you are doing before you do it. "Can I take the opposite side for five minutes, just to stress-test this? I actually like the plan." That sentence tells the room you are helping, not undermining, and makes it easier for others to join in.

2. Steelman first

Before you attack the idea, state it in its strongest form. "So the case is: our audience lives on this platform, this creator has a large following in our niche, and one big push gets more attention than five small ones. Did I get that right?"

This proves you understand the plan, and it keeps you honest: you are now attacking the real argument, not a weaker version. The full technique is covered in our guide to steelmanning.

3. Attack the plan, not the person

Talk about the idea, the numbers and the assumptions. Never about the person's judgment or motives.

4. Go after assumptions and failure modes

The most useful devil's advocate does not argue that the whole plan is bad. They find the two or three assumptions the plan rests on and ask what happens if each one is false. A plan built on "customers will pay $20 a month" and "we can ship by March" has two load-bearing beams. Kick those, not the paint.

5. Time-box it

Give devil's advocacy a fixed slot: ten minutes in a meeting, one page in a proposal, one evening for a personal decision. Then stop and decide. The goal is a better decision, not an endless one.

6. Rotate the role

If the same person always plays devil's advocate, they get labeled as negative and everyone else gets lazy about criticism. Rotate it. The quiet person with unvoiced doubts may be your best devil's advocate once they have official permission.

7. End with a verdict

Close explicitly. "We found three risks. We'll accept the first, fix the second with a pilot, and the third is a dealbreaker unless we get a signed contract."

If you want to warm up with a question that splits almost every team, try playing the other side of this one:

25 devil's advocate questions to ask

You do not need to be clever to be a good devil's advocate. You need good questions. Here is a bank you can copy, grouped by what they test.

Assumptions

1. What has to be true for this plan to work?
2. Which of those things are we least sure about?
3. What are we assuming about the customer (or user, or reader) that we have not checked?
4. If we had to bet our own money on this assumption, would we?
5. What would change our minds?

Risks

6. Imagine it is a year from now and this failed. What is the most likely reason?
7. What is the worst realistic outcome, and could we survive it?
8. What is the one thing that, if it went wrong, would sink everything else?
9. How easy is this to reverse if we are wrong?
10. Who is going to be most unhappy about this, and what can they do about it?

Alternatives

11. What are we giving up by choosing this?
12. What would we do if this option were not available?
13. Is there a smaller, cheaper version we could test first?
14. What would a competitor, or a smart outsider, do instead?
15. Why now rather than in six months?

Evidence

16. How do we know that?
17. Where did this number come from, and how old is it?
18. Are we looking at the successes and ignoring the failures?
19. Would this evidence convince a skeptic who did not already want it to be true?
20. What evidence would we expect to see if we were wrong, and have we looked for it?

Second-order effects

21. If this works, what problems does success create?
22. How will people change their behavior in response to this?
23. What does this commit us to next year?
24. Who else is affected who is not in this room?
25. If everyone did this, what would happen?

Pick three or four per decision, not all twenty-five. For the evidence questions, it helps to know the common reasoning traps; our guide to logical fallacies is a good companion, and Socratic questioning goes deeper on asking questions that reveal gaps without attacking anyone.

Scripts you can use word for word

Knowing you should push back is easy. Knowing what to say in the moment is the hard part. Here are scripts for three common situations.

In a meeting

Opening the role:

"Before we lock this in, can I play devil's advocate for five minutes? I'm mostly on board. I just want to make sure we've tested the weakest parts."

Raising the objection:

"The part I keep coming back to is the assumption that customers will switch within the first quarter. If that takes six months instead of three, what does the plan look like?"

Inviting others:

"Is anyone else holding a doubt they haven't said yet? This is the moment for it."

Closing:

"Okay, that's my devil's advocate hat off. I think the plan survives, as long as we add a checkpoint at month two."

For your own decisions

Playing devil's advocate against yourself is harder, because you are both the defender and the attacker. Write it down; arguing in your head tends to go in circles.

"My plan is to [decision]. The strongest case for it is [steelman your own view].

Now the case against: If I'm wrong, the most likely reason is ______. The assumption I'm least sure of is ______. The person who would disagree with me most strongly would say ______. The option I haven't seriously considered is ______.

After writing this, I'm still going ahead / I want to test ______ first / I've changed my mind."

Amara used this template before accepting a job with a bigger title and a longer commute. The line that stopped her was the one about her strongest critic. She wrote: "My partner would say I complained about a 40-minute commute for two years and this one is an hour." She did not turn the offer down. She negotiated two remote days before signing: not a veto, a better version of the same decision.

For more structured ways to pressure-test personal choices, see how to make better decisions with AI.

For students

When a teacher assigns you a devil's advocate position, argue it as its smartest supporter would, not as a cartoon.

"I'm arguing the other side today, so this isn't necessarily my view. The strongest argument for this position is ______. The evidence its supporters would point to is ______. Where I think the usual view is weakest is ______."

In an essay, add a "strongest objection" paragraph before your conclusion: state the best counterargument fairly, then show why your thesis still holds or how you would qualify it.

How to receive devil's advocate pushback

When someone attacks your plan, even politely, your instinct is to defend it. Resist that for a minute.

• Thank them first. "Good, that's exactly the kind of thing I want to hear now rather than in March." It costs nothing and it keeps the doubts coming.
• Separate the objection from the tone. A clumsy objection can still be correct. Ask yourself: if a person I really respect had said this, would I take it seriously?
• Sort objections into three piles. Fatal (the plan does not work if this is true), fixable (adjust the plan) and accepted risk (we know, we are doing it anyway). Most objections are fixable.
• Ask for specifics. "What would you need to see to feel good about this?" turns a vague doubt into a test you can run.
• Do not rebut in real time. If you argue every objection as it lands, people stop raising them. Write them down and respond at the end.
• Change the plan visibly. If a point improves the plan, say so and credit the person.

And if the pushback shows the plan should die, the process just saved you money.

A worked example: Tobias's second location

Here is the whole method on one real-sized decision, so you can see it in motion.

The decision: Tobias owns a busy neighborhood coffee shop. A storefront has opened up across town, and the landlord wants an answer in two weeks. Tobias is excited. His business partner, Rosa, agrees to play devil's advocate for one evening before they decide. Their accountant, Amara, joins for the numbers.

Step 1: Rosa steelmans the plan

"Here's the case as I understand it. The first shop has had a waiting line on weekends for a year, so demand is proven. The new area has offices but no good independent café. We already have a supplier, a brand and recipes, so the second shop costs less to launch than the first. And the lease rate is lower than what we pay now. Fair?"

Tobias says yes.

Step 2: Rosa finds the load-bearing assumptions

1. The weekend line means there is enough demand for a second shop.
2. The offices nearby will produce morning traffic every weekday.
3. Tobias can run two shops without the first one getting worse.

Step 3: She attacks each one

On demand:

"The line is at our shop, in our neighborhood. Those customers aren't moving across town. The question isn't whether people like our coffee. It's whether people over there will, and we have no evidence about them yet."

On the offices:

"How many of those offices are full five days a week? If they're mostly hybrid, the morning rush might be Tuesday to Thursday, which is a very different revenue line."

On Tobias himself:

"Right now you're in the shop six days a week, and that's a big part of why it works. Who is the version of you at the new location?"

Step 4: Amara runs the numbers

Amara models three scenarios: the plan as imagined, only three strong weekdays, and the original shop dipping while Tobias splits his time. None is a disaster alone, but in the combined worst case the new shop takes uncomfortably long to cover its costs.

Step 5: The verdict

They sort Rosa's objections:

• Demand across town: fixable. Tobias will run a pop-up cart near the new location on four weekday mornings before signing.
• Hybrid offices: fixable. Tobias will count foot traffic outside the storefront on different days and ask the landlord for occupancy details.
• Tobias spread too thin: potentially fatal. They agree the plan only works if they hire and train a shop manager for the original location first.

The landlord grants a three-week extension. The pop-up does well midweek and poorly on Mondays and Fridays. They sign, but with a lower first-year target and a manager hired before opening.

Nobody "won." The plan survived, changed shape, and got safer. To lay out both sides of your own version quickly, try the AI pros and cons tool before bringing in a human critic.

Here is a close cousin of Tobias's question, worth arguing both ways:

Using an AI devil's advocate

The biggest problem with devil's advocacy is social. Colleagues do not want to seem negative, friends do not want to hurt your feelings, and you go easy on yourself. An AI will attack your favorite idea at 11 p.m. without worrying about Monday's team lunch. On Debatly, you have a few options:

• The Devil's Advocate debate format. Enter your position or plan, and one character argues against it across several rounds while you defend. A judge such as Judge Alex then weighs the arguments and explains the verdict. You can start one in the debate room.
• Chat with Petra, the devil's advocate. Petra is built to find the weakest link in whatever you bring her. Paste in your plan and ask her to steelman it first, then attack the top three assumptions.
• Otto, the contrarian, takes the opposite side of almost anything, useful when you suspect you are surrounded by agreement.
• Max, the risk analyst, focuses on what could go wrong, how likely it is and how bad it would be: a one-person premortem.

You can also bring several of them into a group chat and let them argue about your plan with each other. For bigger choices, the AI decision maker pulls this into a structured process.

An honest limit: an AI devil's advocate only knows what you tell it. It cannot see that your co-founder is burned out, that your customers hate phone calls, or that the storefront floods every spring. It is also an assigned dissenter by design, so Nemeth's point applies: treat it as a way to generate objections, then test the best ones with people who know your situation and actually disagree with you.

One more to try, this time as a pitch you have to defend:

Make devil's advocate a habit, not a personality

A helpful devil's advocate differs from an annoying one in how they push, not how hard. Announce the role. Steelman first. Go after assumptions, not people. Keep it short, rotate it, and end with a decision.

Pick one decision you are facing this week, big or small, and run it through the "for your own decisions" script above. Then, if you want an opponent who will not go easy on you, open a Devil's Advocate debate on Debatly and let Petra take her best shot. Try for free, and bring the plan you are most sure about.