A room full of smart people can still make a decision that looks baffling in hindsight. Everyone had experience, the meeting felt productive, nobody raised a serious objection—and six months later, the team is asking how an obvious warning sign slipped past them.
That uncomfortable gap between individual intelligence and collective judgment is where groupthink becomes useful. It is not proof that collaboration does not work. It is a reminder that good group decisions require more than talented people; they require a process that makes disagreement, unusual information, and uncertainty difficult to suppress.
What is Groupthink
Psychologist Irving Janis popularized the term groupthink to describe situations where the desire for agreement becomes strong enough to interfere with sound decision-making.
That makes groupthink particularly relevant to modern workplaces. Product launches, security decisions, hiring panels, investment committees, medical teams, boards, and AI strategy groups routinely depend on people combining specialized information that no single member possesses.
The First Problem: Groups Repeat What Everyone Already Knows
One of the strangest findings in group-decision research is that groups often spend too much time discussing shared information—facts already known by several members—and too little time discussing information known by only one person.
Researchers call this the hidden-profile problem. In classic experiments, groups were given information distributed across members so that the best choice became obvious only if everyone pooled their unique knowledge; groups often failed to do so and therefore missed the superior option.
A striking example from the foundational research illustrates how serious this can become. When all group members received the full information, 83% of groups selected the best option; when crucial information was divided among members, only 18% did.
Imagine a software team evaluating a vendor. Everybody knows the price is excellent, but only one engineer knows the platform creates a major security problem. If the meeting spends 40 minutes celebrating the price and two minutes on the security concern, the team technically had enough knowledge to make the right decision—it simply failed to use it.
The Second Problem: Silence Starts Looking Like Agreement
Once a majority preference begins forming, something subtle happens. People with doubts may soften them because they assume everyone else has information they do not, or because challenging a confident group feels socially expensive.
Then silence gets interpreted as consent. Apparent consensus makes the group even more confident, which makes dissent feel increasingly awkward.
This is why asking, “Does everyone agree?” is often a poor decision tool. By that point, the meeting has already revealed the socially preferred answer.
Try a better question: “What information would make this decision wrong?” It invites evidence rather than forcing someone to announce that the entire room is mistaken.
Hierarchy Can Distort the Discussion Before It Begins
A senior leader does not need to say, “Do not disagree with me” to influence a room. Simply announcing a strong preference early can provide an anchor around which everyone else's analysis begins organizing itself.
I have found a simple meeting rule surprisingly useful: the most senior decision-maker gives their recommendation last. Junior members, specialists, and people closest to the operational details get to reveal what they know before learning which answer carries the most organizational weight.
This does not remove authority. The leader still makes the decision when appropriate; it simply prevents leadership from accidentally deciding which evidence feels welcome before the discussion has properly started.
You can strengthen the effect by asking everyone to record an initial assessment privately before the meeting. That preserves independent thinking long enough to compare it with the group's eventual conclusion.
Smart Teams Can Become Overconfident Together
Agreement feels reassuring because repetition can resemble confirmation. If five people mention the same positive fact, the group may experience that information as increasingly important—even if all five learned it from the same source.
The number of people repeating an idea is not the same as the number of independent pieces of evidence supporting it.
That distinction is especially important in technology. Ten employees may all believe a feature is popular because they saw the same dashboard, while one customer-support analyst has direct evidence that users misunderstand it. The repeated metric sounds stronger socially; the lonely piece of contradictory information may actually be more informative.
Before approving a major decision, ask: How many genuinely independent signals support this conclusion? It is a small question that can puncture surprisingly large bubbles of confidence.
Design Meetings to Find Information Before Defending Opinions
Research on hidden profiles suggests that an advocacy-heavy discussion can interfere with discovering the best option. Groups operating more like impartial fact-finders can have a better chance of surfacing unshared information before preferences take over.
That leads to a cleaner meeting structure. Separate information collection from recommendation instead of mixing them from minute one.
A practical sequence looks like this:
- First, each person shares information others may not have.
- Next, identify assumptions and areas of uncertainty.
- Then compare at least two realistic alternatives.
- Only afterward should people advocate for their preferred option.
- Before deciding, name what evidence would justify reversing the decision later.
That last step matters because a good decision process should make updating easier. If the group has already defined the evidence that would change its mind, new information becomes a trigger for reassessment rather than an attack on everybody's credibility.
Do Not Confuse Psychological Safety With Endless Niceness
Teams often say they encourage disagreement while quietly rewarding people who make meetings smoother. Real psychological safety becomes visible when somebody challenges a popular idea and the group stays curious instead of becoming defensive.
Watch what happens after the first uncomfortable question. Does somebody investigate it, or does the room immediately explain why it is not relevant?
Competition can make information-sharing even more difficult. Experiments using hidden-profile tasks found that competitive conditions encouraged people to withhold unique information and made them less willing to challenge their initial preferences compared with more cooperative settings.
The solution is not creating a workplace where nobody competes. It is making it clear that during high-stakes decisions, surfacing information that improves the outcome matters more than protecting personal territory.
Make Dissent a Role Everyone Can Use
“Playing devil's advocate” has become a meeting cliché because organizations know dissent is useful. The stronger approach is to make critical testing part of everyone's job rather than relying on one designated contrarian.
Before final approval, run three short tests: ask what the team might be assuming without evidence, identify the strongest argument for the rejected alternative, and imagine the decision has failed one year from now.
That final exercise is often called a premortem: instead of asking whether the plan will fail, assume it already did and work backward to identify plausible causes. It makes criticism less personal because people are examining a fictional future failure rather than directly attacking a colleague's proposal.
Technology Can Help—or Industrialize the Problem
Digital collaboration tools can make independent input easier. Anonymous polls, asynchronous idea collection, shared documents, and pre-meeting questionnaires can capture perspectives before the loudest or most senior voice establishes the room's direction.
But technology can also amplify conformity. Recommendation systems, dashboards, shared AI summaries, and automated scoring can give everyone the same information source, creating an illusion of independent agreement when the underlying evidence is identical.
That makes source diversity increasingly important. If an entire decision team is asking the same AI model, reading the same dashboard, or relying on the same analyst note, you may have ten opinions but only one informational bloodstream.
Direct Answers
Groupthink is a process problem, not an intelligence problem. Smart people can make poor decisions when maintaining consensus becomes more important than testing evidence.
Ask for unique information before opinions. Groups naturally over-discuss facts everybody already knows, so deliberately surface what only one or two people know.
Let powerful voices speak later. Early leadership preferences can anchor discussion before alternative interpretations receive a fair hearing.
Count independent evidence, not repeated agreement. Five people citing the same source do not provide five separate confirmations.
Predefine what could change the decision. Naming reversal criteria before commitment makes future updating look responsible rather than embarrassing.
The Best Teams Make Agreement Earn Its Place
Strong teams do not avoid consensus. They simply refuse to rush into it.
They collect information before defending positions, treat unusual evidence as valuable rather than inconvenient, let hierarchy guide decisions without dominating discovery, and make changing one's mind a sign of better judgment rather than weakness.
That may make some meetings slightly less comfortable. Good.
If everyone agrees immediately on an important, complicated decision, the smartest next question may not be “Can we move on?” It may be “What haven't we heard yet?”
