Most companies measure psychological safety once a year, with one question buried inside an engagement survey. HR reports the score and the cycle closes. The number goes in a deck.
That score cannot tell you what someone does when they have bad news and a good reason to stay quiet. Psychological safety has one job. It moves accurate information to the person who can act on it, early enough to matter.
Gallup’s State of the Global Workplace 2026 report put global employee engagement at 20% in 2025, the lowest reading since 2020. Whatever your last survey said, some of your people have already decided what is safe to tell you.
Two large studies point in the same direction. Google’s Project Aristotle studied 180 internal teams. It ranked psychological safety first among five dynamics separating the most effective ones. A field experiment published by CEPR in 2025 randomly assigned more than 1,000 teams at a global healthcare company to different instructions for manager one-on-ones. Teams whose managers focused on people’s psychological needs got measurably safer. Teams whose managers focused on task execution improved only from a moderate starting point.
Boeing has a speak-up program
Boeing runs a formal employee reporting channel called Speak Up. It has existed for years, and it did not move the information that mattered. Kursten Faller and Alan Weiss covered the Boeing case on this site in June, in “Silent Killers of Project Success”.
In 2024, an expert panel convened for the Federal Aviation Administration found employees still hesitated to report safety concerns. Managers investigated reports inside their own chain of command. Interviews surfaced a second problem. Employees distrusted Speak Up’s anonymity, and many preferred to avoid the formal systems and go to their manager instead.
Boeing had run its own survey in 2016, during 737 Max certification. Of the authorized representatives who responded, 39% perceived undue pressure. Twenty-nine percent worried about the consequences of reporting it. A whistleblower gave that survey to Congress, which published it in 2020.
Two 737 Max crashes, Lion Air Flight 610 in October 2018 and Ethiopian Airlines Flight 302 in March 2019, killed 346 people.
Harvard Business School professor Amy Edmondson wrote that the crisis was “a textbook case of how the absence of psychological safety … can lead to disastrous results.” The channel existed the whole time. Boeing’s own 2016 survey already showed people were holding back, eight years before the FAA panel found the same thing.
Most psychological safety programs stop at stage one
Timothy R. Clark, founder and CEO of LeaderFactor, breaks psychological safety into four stages. Inclusion safety: can I be myself here. Learner safety: can I ask a question or get something wrong. Contributor safety: does my judgment count. Challenger safety: can I tell you that you are wrong.
Across more than 500 founders and executives I have coached, almost all the work lands on stage one. Inclusion is the stage with a budget line, an onboarding module, a hashtag and a slide in the all-hands meeting. Challenger safety gets none of that. Challenger safety is the stage where an engineer tells a vice president the plane is not ready to fly.
I once worked at a company that branded itself around belonging. There were panels, inclusion language on the walls, a values page and real pride in how open the place was. I spent months there feeling worthless, and eventually they let me go. I emailed my manager afterward and asked her, in good faith, what I could have done differently, and she wrote back telling me to take a job I actually liked next time.
Not one question about the room she ran. That company would have scored well on any inclusion survey you handed it.
Which stage your team is at
Set the survey aside and answer four questions from your last 10 meetings.
Did a new hire speak without being called on? That is stage one. Did someone say, “I don’t understand this,” with no apology attached? Stage two. Did a junior person’s idea change a decision that had already been made? Stage three. Did anyone tell you that you were wrong, out loud, with peers in the room? Stage four.
The stages stack. Your real level is the highest one you can name a specific example for from the last 90 days.
Five behavioral measures of psychological safety
Three of these run on data you already collect. Two cost an hour of your own time.
1. Track your time-to-bad-news
Take your last three incidents. For each one, find out when the first person knew something was wrong, then find out when it reached the person who could actually fix it, and write down the days between those two moments. That gap is your number. Run it every quarter and log the trend.
2. Count disagreements by rank
Sit in your own meetings and count one thing. How often does someone disagree with the highest-paid person in the room? Log it for 30 days. Record the count each week, including the weeks it comes back zero.
3. Stop reading your error-reporting rate as a safety score
A 2015 study in PLOS One, led by Ann-Marie Howell, analyzed 5.8 million incident reports from England’s National Reporting and Learning System over a decade. The authors found no link between how much a hospital reported and its mortality or patient satisfaction. Hospitals that reported more had fewer litigation claims and better safety-culture scores. In their words, a high volume of incident reports “does not signify an error-prone environment.” So read the count as a measure of how safe people feel filing one. When trust improves, expect it to climb. If nothing gets filed for a whole quarter, ask why in your next staff meeting.
4. Code your exit interviews by stage
You collect this data already and mostly waste it. Tag every exit comment to a stage. “I never fit in here” is inclusion. “I got burned for asking” is learner. “I was micromanaged” is contributor. “My ideas went nowhere” is challenger. Two quarters of tagged exits will show you where the culture breaks, in your people’s words instead of HR’s. It also catches the quiet cracking pattern while the person is still on payroll.
5. Ask who pushed back last
Ask five people to name the last person who challenged a senior decision out loud. If three name the same person, and all three know how it ended for them, you have your answer. Run this with people who do not report to you, or you will get the answer they think you want.
Where to start this quarter
Pick one of the five. Time-to-bad-news is the easiest to build, since it comes out of incident reviews you already run. Give it two quarters before you decide what the number means.
Start counting disagreements in your next staff meeting. You will have a number by Friday.
Post Views: 322
