The overnight technician knew. That was the part that stayed with me.

Years ago, running operations at a large telecom, I sat in a change review for a maintenance window on a core site. Eight people on the call. We walked through the rollback plan, the customer notification, the staffing, the timing. Everyone had read the same change ticket, so everyone talked about the same change ticket, and the window was approved in about twenty minutes. It went badly. The site’s backup power had a known fault that had been logged by field staff two weeks earlier, and when the planned cutover dropped primary power longer than expected, there was nothing underneath it.

The technician who logged that fault was on the call. When I asked him afterward why he hadn’t raised it, he said something I’ve heard in different words many times since: “Nobody asked about power. I figured if it mattered, it’d be on the ticket.”

He wasn’t hiding anything. The meeting simply never created a moment where the thing only he knew had a place to land. Psychologists have a name for that pattern, and four decades of research on it should change how you run every decision meeting on your team.

What researchers found when they split the facts up

In 1985, Garold Stasser and William Titus ran an experiment that looked almost too simple to matter. Small groups of students had to pick the best of three candidates for student body president. Each person got a sheet of facts about the candidates before the discussion. In some groups, everyone got the full set. In others, the facts were split so that the information favouring the best candidate was spread across different people, while the information favouring a weaker candidate was handed to everyone.

If the groups had pooled what they knew, the best candidate would have been obvious. They mostly didn’t. Discussion gravitated toward the facts everyone already had, and the groups tended to choose the candidate their members preferred going in rather than the one the full record supported. Stasser and Titus called this a hidden profile: the right answer exists inside the group, but no single member can see it alone, and the conversation never assembles it.

You might hope that was a quirk of undergraduates and a lab. It wasn’t. A meta-analysis of 65 hidden profile studies covering more than 3,000 groups found that groups facing a hidden profile were eight times less likely to find the correct solution than groups where everyone had the full picture. The same analysis found that groups mentioned far more of their common information than their unique information, by roughly two standard deviations.

The broader workplace evidence points the same way. Jessica Mesmer-Magnus and Leslie DeChurch pooled 22 years of research covering about 4,800 groups and more than 17,000 people and found that sharing information predicted team performance, but the kind of sharing mattered more than the amount. As Mesmer-Magnus put it in the American Psychological Association’s summary of the work, “it’s more important what the teams are talking about, than how much they are talking.” Most teams, they found, spent their time on information that the rest of the group already knew.

Why the shared facts win the airtime

None of this requires anyone to be lazy, political or afraid. Three ordinary forces do the work.

Repetition is built into the math

If five people know a fact, five people have a chance to mention it. If one person knows a fact, one person does. Before anybody’s psychology enters the picture, shared information simply has more chances to come up, and once it comes up, others recognise it and add to it.

Shared facts feel safer to say

When you mention something everybody knows, heads nod. It confirms you’re competent and on the same page. When you mention something nobody else has heard, you’re out on your own. No one can back you up, and some people will quietly wonder whether you’ve got it right. The technician on my call had a fact no one could confirm, about a topic nobody had raised, in a meeting where the senior people were all agreeing. The social cost of speaking was higher than the visible benefit.

People argue for the answer they walked in with

This is the force managers underestimate most. A team of German researchers led by Nadira Faulmüller and Stefan Schulz-Hardt tested what happens when you remove the conversation problem entirely and just hand people all the information. In their experiment, every group given the full picture up front found the correct answer, while only 1 of 15 groups working from split information did. More striking, they estimated that even when all the relevant facts get exchanged, almost half of groups would still miss the right answer, because members judge new facts through the preference they formed before the meeting. Information that supports your early view feels credible. Information that cuts against it feels like noise.

That last finding is why “just make sure everyone shares” isn’t enough. The order in which you collect opinions and facts matters as much as whether the facts come out at all.

Where it shows up on a real team

Once you know the pattern, you’ll see it everywhere. A few places I’ve watched it cost real money:

• Hiring debriefs. Every interviewer saw the same résumé, so the debrief turns into a discussion of the résumé. The one interviewer who noticed the candidate dodged every question about a failed project mentions it late, gently, and it gets weighed against six people’s enthusiasm.
• Vendor selections. The demo was shared, so the demo dominates. The person who called a reference and heard about a six-month implementation slip has one data point against a room full of slides.
• Project go/no-go calls. The status report is shared. What the engineer who’s been fighting the integration for three weeks knows about its real state is not.
• Incident reviews. The timeline is shared. The fact that the alert had been firing intermittently for a month, known to whoever was on night shift, often isn’t.

In every case the meeting feels thorough. People talked, everyone contributed, the decision was discussed. That feeling is exactly the trap. A discussion made of shared facts feels complete because nobody in it knows what’s missing.

The leader’s job is to change what gets repeated

Here’s the finding I’d want every new manager to know. In a 1996 study of three-person physician teams diagnosing medical cases, James Larson and colleagues found the same bias toward shared information. But team leaders behaved differently from other members: they repeated substantially more case information, and over the course of the discussion they repeated unshared information at a steadily increasing rate. A follow-up study found that pooling unshared information, and not shared information, was what improved the accuracy of the teams’ diagnoses.

That gives you a concrete role. When someone says the thing nobody else knew, your job is to pick it up and put it back on the table, more than once. Repetition from the person running the meeting tells everyone that the fact counts. If you let it pass with a “good point” and move on, the group will return to the facts it already shares within a minute.

Some specific ways to run decisions so the hidden facts surface:

Collect facts before you collect opinions

The worst opening line in a decision meeting is “So what does everyone think?” It invites people to declare a preference, and once they have, the research above says they’ll filter everything else through it. Open instead with “What do you know about this that might not be in the document?” and hold the vote, or the go-around, until that question has been answered by everyone.

A practical version I’ve used for years: send the question before the meeting and ask for written replies. Writing removes the social cost of being the lone voice, and it stops the first confident speaker from setting the frame for everyone else.

Say out loud who knows what

Stasser and two colleagues later tested whether it helps to tell groups which member holds extra information on which topic. In their study of groups solving a murder mystery, publicly assigning expert roles increased the discussion of unshared clues and helped groups pick the right suspect. Privately telling people they held extra information didn’t produce the same benefit; the rest of the group had to know too.

On a real team this is simple. “Before we decide, I want to hear from Field Ops on site condition, from Finance on the contract terms and from the on-call lead on what the last month of alerts looked like.” You’ve just told the room that those three people hold something the others don’t, and you’ve given each of them permission to speak to it.

Frame the goal as getting it right, not getting agreement

Mesmer-Magnus and DeChurch found teams shared more useful information when they were told to find a correct or best answer rather than a consensus. Consensus framing rewards convergence, which rewards the shared facts. Accuracy framing rewards the oddball fact that changes the answer. Saying “I don’t need us to agree, I need us to be right” at the start of a decision meeting is not a throwaway line. It changes what people think the meeting is for.

Protect disagreement when it shows up

Work led by Stefan Schulz-Hardt found that genuine disagreement among members before discussion improved the quality of group decisions in hidden profile situations. Disagreement keeps the conversation open longer and pushes people to examine information instead of settling early. If your team always agrees quickly, that isn’t a sign of alignment. It may be a sign that the meeting ends before the unshared facts arrive. I wrote about a related failure, where everyone agrees to a plan nobody actually wants, in the piece on the Abilene paradox.

Ask the last question directly

Before you close any decision that matters, ask: “Is there anything any of you knows that would change this, even if you’re not sure it’s relevant?” Then wait longer than feels comfortable. That question would have saved my maintenance window. The technician told me later he would have answered it.

When this is worth the effort

Not every decision deserves this treatment. Picking a date for the team lunch doesn’t need a structured fact-gathering round. The hidden profile risk is highest when three conditions overlap:

1. The people in the room have genuinely different vantage points: different roles, shifts, customers or systems.
2. There’s a shared document or presentation everyone has read, which will naturally dominate the discussion.
3. The decision is costly or slow to undo.

When all three are true, assume the decisive fact is sitting with one person and build the meeting to find it. If you want a second safeguard for big calls, a premortem does similar work from a different direction, by giving people explicit permission to say what might go wrong.

The technician on that change review had been with the company longer than most of the managers on the call. He wasn’t timid and he wasn’t careless. He was doing what the meeting taught him to do: respond to what was asked. Most of the people on your team are doing the same. If the questions you ask only cover the facts everyone already has, that’s all you’ll ever hear about, and you’ll make decisions with confidence that the room never earned.