Why does your ‘aligned’ team keep making bad decisions?
Irving Janis published his groupthink theory in 1972 after studying foreign policy disasters: Bay of Pigs, Pearl Harbor, Vietnam. His finding: highly cohesive groups suppress dissent, overestimate their own competence, and rationalize warning signs. The stronger the group identity, the stronger the pressure to conform.
Janis wasn’t writing about product teams.
The 8 Symptoms, Translated
Illusion of invulnerability. “Our competitor can’t catch us. We have a two-year head start.” Excessive optimism that blinds the team to risk.
Collective rationalization. “The beta users complained about onboarding, but they’re not our target persona.” Dismissing contradictory evidence.
Belief in inherent morality. “We’re making the world more open and connected.” Assuming the mission justifies any decision.
Stereotyping outsiders. “They just don’t get the vision.” Dismissing critics as uninformed rather than informative.
Direct pressure on dissenters. “Are you not a team player?” The PM who questions the roadmap gets sidelined.
Self-censorship. The PM who has doubts stays quiet because everyone else seems aligned. Silence mistaken for agreement.
Illusion of unanimity. “Everyone’s on board.” Nobody objected, so there must be consensus. Silence isn’t consent.
Mindguards. Certain team members protect the group from contrary information. “I wouldn’t bring that up, it’ll derail the conversation.”
Score your last decision session against all eight. How present was each symptom in the room?
Think about your last decision session and score the room as a whole on the following. How present was each of these statements in the room?
Scale: 0 = not present in the room · 5 = strongly present.
- Illusion of invulnerability"We've done this before, we know what we're doing."
- Collective rationalization"That data doesn't apply to our situation."
- Belief in inherent morality"We're the good guys, we wouldn't do that."
- Stereotyped out-groups"The other team just doesn't get it."
- Self-censorship"Everyone looks aligned, I'll hold this back."
- Illusion of unanimity"Nobody objected, so we're agreed."
- Direct pressure on dissenters"Let's not open that can of worms."
- Self-appointed mind-guardsSomeone shields the leader from bad news.
Answer the above and select “How bad was it?” to see your groupthink risk on a scale of 0–40.
This is an illustrative diagnostic based on Janis (1972). It describes the room, not the people in it. Use it to design the next session, not to blame anyone.
The “Strong Culture” Trap
Strong product cultures are especially vulnerable. The things that make a team effective (shared values, trust, identity) also make dissent costly. When everyone believes in the mission, questioning it feels like betrayal.
“Disagree and commit” became a mantra, but most orgs only do the second half.
Antidotes That Actually Work
Janis proposed structural fixes, not cultural ones. They translate cleanly:
- Assign a devil’s advocate, and rotate the role. One person per planning cycle whose job is to argue against consensus. Rotate it so nobody gets typecast.
- Bring in an outside voice. Someone from a different team or domain. Show them the plan and ask: “What are we missing?”
- Separate generation from evaluation. Don’t brainstorm solutions and evaluate them in the same meeting.
- Make dissent structured. A prioritization matrix that forces explicit tradeoffs is an anti-groupthink device. Groupthink hates forced choices.
Forced choice is the anti-groupthink device, and the demo is where it bites. When a stakeholder sees working software that does not match their expectation, silence becomes unsustainable. They either speak up in front of everyone or the decision stands. Scope creep falls not because anyone says no harder, but because disagreement can no longer stay silent.
Groupthink and AI Agents
AI agents introduce a new risk: the agent is the ultimate conformist. It doesn’t have dissenting instincts. It won’t push back on a bad plan unless explicitly prompted. A human team with groupthink tendencies, augmented by agents that amplify consensus, is a decision-making disaster waiting.
The fix: instrument agents to surface uncertainty and alternatives. “Here’s the proposal. Now give me three reasons it might fail.” If your agent can’t do that, it’s part of the problem.
Your team might be aligned. It might also be in groupthink. The difference is invisible from inside, which is why the structural antidotes matter: rotate the devil’s advocate, force the tradeoff, and let an outside voice in. s@spencervaradi.com.
Related: Your Sprint Retro Is Lying to You.