The Workplace Wellness Report

The randomised evidence

Start with the number that built an industry. In 2010, Katherine Baicker, David Cutler and Zirui Song published a meta-analysis in Health Affairs and reported that medical costs fell about $3.27 for every dollar spent on workplace wellness programs, and absenteeism costs fell about $2.73. Those two figures went into a decade of vendor decks, benefits committee slides and board memos. You have almost certainly seen them without being told where they came from. What the authors wrote was more careful than how the number travelled. They were summarising the literature that existed at the time, and that literature was almost entirely observational.

Observational means nobody was randomised. An employer offered a program, some employees signed up, and researchers compared the joiners to everyone else. The problem with that comparison is not subtle. The people who sign up for the step challenge are not a random slice of your workforce. They are the ones who sleep, who have childcare, who are not working a second job, who already have a doctor. Compare them to everybody else, find that they cost less, and you have measured who they already were. You have not measured what the program did. Economists call it selection, and it is the single biggest reason the wellness return-on-investment literature looked so good for so long.

The first serious crack came in 2013. RAND published the Workplace Wellness Programs Study, commissioned by the Departments of Labor and Health and Human Services. Soeren Mattke and colleagues paired five years of medical claims with program data. They did find statistically significant, clinically meaningful improvements among participants in exercise frequency, smoking behaviour and weight control, though not in cholesterol control. On money, the finding was quieter than anyone wanted. Participation over five years was associated with an average annual difference of about $157 in health care costs, and that difference was not statistically significant. Read it again. Not significant. The most-cited wellness evaluation of its decade could not locate the savings.

Then two teams did the thing nobody had done at scale. They randomised. Damon Jones, David Molitor and Julian Reif ran the Illinois Workplace Wellness Study at a large employer with more than 12,000 workers, and they assigned program eligibility and financial incentives at the individual level. Not by worksite. Not by volunteer. By lottery. More than 56 percent of eligible workers took part in the first year, which is a participation rate most employers would envy. The analysis sample for health outcomes was 4,834 employees, 3,300 of them eligible for the program and 1,534 in the control group, followed for two years.

Here is what the lottery revealed. The authors found strong patterns of selection: the people who chose to participate already had lower prior medical costs and healthier baseline behaviours than those who did not. That is the bias, caught in the act, in the same study that could then subtract it out. And once it was subtracted out, the effects went away. In year one there were no significant causal effects on total medical spending, on health behaviours, on productivity or on self-reported health. The confidence intervals were tight enough to rule out 83 percent of previous published estimates on medical spending and absenteeism. Not merely unproven. Largely excluded.

The two-year results, published in JAMA Internal Medicine in 2020, are worth quoting precisely because they are so flat. Across sixteen biometric measures there was nothing: body mass index moved minus 0.40 with a confidence interval from minus 1.11 to 0.31, systolic blood pressure minus 0.15 millimetres of mercury from minus 1.48 to 1.18. No effect on diagnoses of diabetes, hypertension or hyperlipidaemia. No effect on office visits, inpatient visits or emergency department visits. Two years, a well-run comprehensive program, real incentives, 64 percent of the treatment group completing at least one component, and the clinical needle did not move.

Report what they did find, because something did. The share of employees who said they had a primary care physician rose from 86.1 percent in the control group to 92.2 percent in the treatment group, a 6.1 percentage point difference with a confidence interval from 3.0 to 9.2. Employees' beliefs about their own risk of obesity, high cholesterol, high blood pressure and impaired glucose improved by 0.07 standard deviations. That is a small, real effect on whether people are connected to care and how they see themselves. It is not the same thing as being healthier, and the authors do not pretend it is.

$157 Average annual difference in health care costs associated with five years of wellness participation — not statistically significant — RAND Workplace Wellness Programs Study, 2013

The second trial was larger and it was run by Zirui Song and Katherine Baicker — yes, the same Baicker who co-wrote the 2010 meta-analysis. Published in JAMA in April 2019, it was a cluster randomised trial across 160 worksites of BJ's Wholesale Club covering 32,974 employees: 20 treatment sites, 20 primary control sites, 120 secondary controls. Registered dietitians delivered modules on nutrition, physical activity and stress reduction. After eighteen months, two self-reported outcomes moved. Regular exercise rose 8.3 percentage points, 69.8 percent against 61.9. Active weight management rose 13.6 points, 69.2 percent against 54.7. Those are genuine, and they are not nothing.

Everything else held still. Twenty-seven other self-reported health measures: no significant difference. All ten clinical markers, including cholesterol, blood pressure and body mass index: no significant difference. Health care spending averaged $3,583 at treatment sites against $3,953 at controls, not statistically significant. Absenteeism, 2.5 percent against 2.6 percent. Job performance and tenure, nothing. Song and Baicker followed the same worksites out to three years and published in Health Affairs in 2021. The behaviour gains persisted at roughly the eighteen-month level. The longer follow-up produced no detectable improvement in clinical, economic or employment outcomes.

No significant differences in clinical measures of health, health care spending and utilization, and employment outcomes after 18 months.Song and Baicker, JAMA, April 2019

Pay attention to what Baicker did here, because it is the most admirable thing in this whole file. She published the optimistic meta-analysis in 2010, then helped build the trial that undercut it, then published that too. That is how the method is supposed to work. The failure is not hers. The failure belongs to every deck that has gone on quoting the 2010 number for sixteen years while two randomised trials and a RAND claims analysis sat there saying otherwise.

So what do you do with this on Monday? Stop selling your wellness program as a cost-control strategy, because the best available evidence says it is not one. Stop calling the vendor dashboard an outcome. Participation is not health. Steps are not blood pressure. A 6.1 point rise in people having a primary care physician is a legitimate thing to be proud of, and it is the kind of claim the evidence will actually bear. If you want the program because employees like it, say that. If you want it because it connects people to care, say that and measure that. If you want it because it will cut your claims spend, the randomised trials have already told you the answer, twice, at three-year follow-up, and the answer was no.

• 12,000+ Employees whose program eligibility was randomised in the Illinois Workplace Wellness Study — source
• 83% Share of previous published estimates on medical spending and absenteeism ruled out by the Illinois confidence intervals — source
• 32,974 Employees across 160 randomised worksites in the Song and Baicker trial — source
• 0 of 10 Clinical markers of health that improved significantly after 18 months in that trial — source
• +6.1 pts Rise in employees reporting they have a primary care physician — the Illinois study's largest real effect — source
• 79% Large US firms offering at least one health and wellness promotion program in 2024 — source

None of this says wellness is bad. It says wellness is not medicine and it is not an accounting strategy. The things that moved in these trials were the things programs are good at: getting people to a doctor, nudging a behaviour someone was already half-ready to change, shifting how a person thinks about their own risk. Those are worth having. Build for those, measure those, and stop promising the board a number that two randomised trials have already taken off the table.

Correlation, properly read

The Gallup Q12 is twelve statements an employee rates on a five-point scale. Do I know what is expected of me. Do I have the materials and equipment I need. Did someone recognise my work in the last seven days. Does someone at work care about me as a person. It takes four minutes. It has been administered for three decades, and the meta-analysis behind it is now in its eleventh edition, published in May 2024. That edition pools 736 research studies across 347 organisations, 183,806 business and work units, and 3,354,784 employee responses, from 90 countries and 53 industries. Whatever else you think, that is not a small sample.

Gallup also does something unusual and to its credit. It states that the meta-analysis includes all available Gallup studies, published or unpublished, which means there is no publication bias in the usual sense — nobody quietly filed the disappointing ones. In most literatures that is the first thing you worry about. Here it is handled, and handled by design rather than by statistical correction. Hold on to that, because it means the correlation is real. What it is a correlation between is the part that gets mangled.

The headline figures are median percentage differences between business units in the top quartile on engagement and those in the bottom quartile. Gallup reports absenteeism 78 percent lower, turnover 21 percent lower in high-turnover organisations and 51 percent lower in low-turnover ones, shrinkage 28 percent lower, safety incidents 63 percent lower, and quality defects 32 percent lower. On the upside: customer loyalty 10 percent higher, productivity 14 percent higher on production records and evaluations and 18 percent higher on sales, profitability 23 percent higher, organisational citizenship 22 percent higher, and net thriving employees 70 percent higher. The true-score correlation between engagement and a composite performance measure is 0.49.

• 0.49 True-score correlation between Q12 engagement and composite business-unit performance — source
• 183,806 Business and work units in the 11th edition meta-analysis — source
• 3.35M Employee responses pooled across 90 countries and 53 industries — source
• 23% Median profitability difference, top versus bottom engagement quartile — source
• 29% US employees engaged, August 2026 — against 26 percent in the year 2000 — source
• 70% Minimum share of the variance in team engagement that Gallup attributes to managers — source

Now say the honest part plainly, because Gallup does. The report acknowledges that it does not directly address issues of causality. It notes that separate causal studies suggest engagement and financial performance are reciprocally related, with engagement the stronger predictor of the two. Reciprocally related. That is a careful phrase and it deserves a careful reading. It means the arrow runs both ways. Engagement may lift performance, and performance almost certainly lifts engagement. A unit that is winning is a more pleasant place to be. People in a growing business feel their work matters because it visibly does.

There is no randomised trial here, and there could not easily be one. You cannot lottery three hundred business units into having a good manager and leave the rest with a bad one for two years. So every plausible alternative explanation stays on the table. Reverse causation: profitable units score higher because they are profitable. Confounding by business conditions: the store in the growing suburb has better numbers and happier staff and neither caused the other. Manager assignment: strong managers get sent to units that are already salvageable, or they get the good territories. Common-method variance in the self-report measures. None of these are exotic objections. They are the first four things any reviewer would ask.

Does not directly address issues of causality.Gallup, Q12 Meta-Analysis, 11th edition, 2024

Then there is the quartile framing, which inflates whatever it touches, and does so in two ways at once. First, top quartile against bottom quartile is a comparison of extremes. You are deliberately ignoring the middle half of your own distribution. The gap between the best and worst quarter of anything is always larger than the gap you will actually close. Second, these are percentage differences in rates, not percentage points. Absenteeism 78 percent lower sounds enormous. If the bottom-quartile rate is four percent of days and the top-quartile rate is a bit under one percent, that is a real and valuable difference, and it is three points of attendance, not seventy-eight of anything. Both statements are true. Only one of them belongs in a slide deck.

Here is the fact that should sober everybody in this industry, and it comes from Gallup too. In August 2026, 29 percent of US employees were engaged and 18 percent were actively disengaged. In the year 2000, the figures were 26 percent and 18 percent. Globally in 2025, 20 percent engaged and 16 percent actively disengaged. Twenty-six years, an entire consulting sector, uncountable surveys and action plans and pulse tools, and the US engaged share has moved three points. Gallup notes that best-practice organisations run at roughly 70 percent. So it can be done. It just has not been done at scale by anybody selling the measurement.

I think the Q12 is genuinely useful, and I want to be specific about why. It is a diagnostic. It tells you where you stand and where the variance sits inside your own organisation. The items are not abstractions about culture — they are twelve observable management behaviours. Did you tell her what is expected. Did you get her the equipment. Did you recognise the work this week. Did you have the conversation about where she is going. That is the actionable part, and it is actionable because it is behaviour, not sentiment. Gallup attributes at least 70 percent of the variance in team engagement to managers, which is the same claim from the other direction.

So use the score the way a thermometer gets used. It tells you that you have a fever. It does not tell you what the infection is, and it is not the treatment. If your engagement number goes up after you fixed scheduling and started doing one-to-ones, the number did not fix anything — you did, and the number noticed. And if someone shows you the quartile table as proof that a survey purchase will raise profitability 23 percent, you now know exactly which sentence in Gallup's own methodology to read back to them.

Plain-spoken version. The evidence says engaged teams and good results travel together, reliably, everywhere, at enormous sample size. The evidence does not say how much of that is the engagement doing the work. Behave accordingly: take the measurement seriously, take the quartile math with salt, and put your effort into the twelve behaviours rather than the score they produce.

Turnover economics

Quitting is not a crisis. It is the weather. The Bureau of Labor Statistics tracks it monthly in the Job Openings and Labor Turnover Survey, and in August 2026 there were 3.1 million quits, a quits rate of 1.9 percent. Total separations were 5.1 million, hires were 5.2 million, and job openings stood at 7.1 million. Every month, in a normal economy, roughly two of every hundred employed people walk out voluntarily. Your organisation is not special and neither is mine. The question is never whether people leave. It is which people, and whether anything you did made the difference.

Now the famous number. Gallup says the cost of replacing an individual employee can range from one-half to two times that employee's annual salary, and puts the total cost of voluntary turnover to US businesses at around a trillion dollars a year. I use that figure and I want to be straight with you about what it is. It is a range, not a measurement. It is built by adding up recruiting costs, interviewing time, onboarding, training, the productivity dip while the new person learns, the productivity dip in the team around them, and sometimes overtime and temporary cover. Every one of those components involves assumptions. Nobody randomised anything. A four-fold range from one-half to two times salary is an honest admission that this depends heavily on the role.

So why keep it? Because the direction and the order of magnitude are not in doubt, and even the bottom of the range is brutal. Half a salary is not a rounding error on a position you were already struggling to fill. Treat it as a planning range, be explicit with your board that it is a range, and never present the two-times figure as if it were a receipt. If you want a defensible number for your own organisation, build it from your own recruiting invoices, your own time-to-productivity data and your own vacancy coverage costs. That takes a week and it is worth more than any benchmark.

3.1 million Voluntary quits in the United States in August 2026, a quits rate of 1.9 percent — BLS Job Openings and Labor Turnover Survey, released 29 September 2026

Gallup's two most useful turnover findings are also survey findings, and they are worth more than the dollar estimate. Fifty-two percent of voluntarily exiting employees say their manager or organisation could have done something to prevent them from leaving. And 51 percent say that in the three months before they left, neither their manager nor any other leader spoke with them about their job satisfaction or their future with the organisation. Both are self-reported and both are collected after the fact, which means the first one is a judgement people make with hindsight. The second one is different. The second one is a factual question about whether a conversation happened, and half the time it did not.

Over half of exiting employees say that in the three months before they left, neither their manager nor any other leader spoke with them about their job satisfaction or future with the organisation.Gallup

Here is where the evidence gets harder, and here is where women come into it. Who carries the care? The Bureau of Labor Statistics measures this directly in the American Time Use Survey rather than asking people to guess. In 2023 to 2024, 38.2 million people provided unpaid eldercare, 14 percent of the civilian non-institutional population aged 15 and over. Fifty-five percent of them were women, 45 percent men. On days they gave care, providers spent an average of 3.9 hours doing it. Employed providers spent 2.8 hours on care days. Those not employed spent 4.8 hours. That last comparison is the whole problem in two numbers.

The 2025 time-use data shows the same tilt in the household. On days they did household activities, women spent 2.8 hours and men 2.1. On an average day, 87 percent of women did household activities against 75 percent of men. In households with a child under six, women spent 2.8 hours a day on primary childcare against 1.7 hours for men. On physical care — bathing, feeding, the parts that cannot be scheduled around a meeting — women spent 1.3 hours and men 38 minutes. Separately, the National Alliance for Caregiving and AARP now count 63 million family caregivers in the United States, one in four adults, up roughly 20 million since 2015. The load is growing.

Then look at the labour force itself. In 2025, according to BLS data on the employment characteristics of families, the labour force participation rate for mothers with children under 18 was 73.9 percent. For fathers it was 93.7 percent. Narrow it to parents of children under six and the gap widens rather than closing: mothers 68.0 percent, fathers 95.3 percent. Twenty-seven points. And note the direction of the father figure — men with young children participate at a higher rate than men with older children. The arrival of a small child pushes mothers out of the labour force and pulls fathers further in. That is not a survey impression. That is the national household data.

• 68.0% Labour force participation, mothers with children under 6, 2025 — source
• 95.3% Labour force participation, fathers with children under 6, 2025 — source
• 38.2M People providing unpaid eldercare, 2023-2024 — 55 percent of them women — source
• 1.3 h vs 38 min Daily physical care of young children, women versus men, 2025 — source
• 63 million US family caregivers, one in four adults, up about 20 million since 2015 — source
• 52% Exiting employees who say their manager or organisation could have prevented their departure — source

Be careful about what this does and does not prove. These are descriptive statistics, not a causal chain. They do not show that any particular woman left your company because of caregiving, and they do not establish a percentage of female turnover attributable to care. Nobody randomised the second shift. What they do show, with government-grade data, is that the care burden is unequal, that it is heaviest exactly when children are small and parents are early in their careers, that eldercare is landing on another 38 million people and growing, and that participation in paid work collapses for mothers of young children in a way it does not for fathers. Put those facts beside the fact that half of departing employees had no conversation about their future, and you do not need a causal estimate to know what to go fix.

I spent 22 years in the Air Force, and the thing the military understood that most employers do not is that readiness is about the whole person and the whole household. Nobody deploys well if home is on fire. The corporate version of that insight is unglamorous: the reason she left is more likely to be Tuesdays at three o'clock than anything in your engagement survey. Fix Tuesdays at three.

The Centre Spread

World Changers

People over profit is not a discount you hand out. It is the cheaper way to operate.

You have been handed a false choice. Run a real business, or do real good. Pick one. The people who hand you that choice are usually not doing either one well. Here is what nobody says out loud: a for-profit company is the most reliable engine for funding impact anyone has built. It has revenue. It has a payroll that already clears. It has customers who already trust it. A non-profit has to raise its entire budget again every single year, from scratch, in a market where attention is the scarcest thing there is. A business earns its budget. That does not make business noble. It makes it durable. Durability is exactly what a mission needs and almost never has.

So stop asking whether you are allowed to make money while you change something. Start asking what your company is actually for, who it carries, and whether the people inside it can feel it. That last question is not soft. It is one of the most heavily measured questions in management, and the numbers are not close.

• 18% of U.S. employees say their current job has a purpose they personally believe in — source
• 20% of employees worldwide were engaged at work in 2025 — a second straight annual decline — source
• 23% higher profitability in top-quartile vs bottom-quartile engaged business units — source
• 7% share of all U.S. charitable giving that came from corporations in 2025 — source

Start with purpose, because that is the word everybody uses and almost nobody measures. Gallup surveyed 4,475 U.S. working adults in August 2025 for a study called The Power of Purpose. Only 18 percent described their current job as one that has a purpose they personally believe in. Forty-five percent said they work primarily to collect a paycheck and benefits. Then look at what purpose does once it is present. Half of the employees with strong work purpose were engaged in their jobs, against 9 percent of those with low purpose. Forty-one percent of the strong-purpose group were watching for or actively seeking a new job, against 68 percent of the low-purpose group. Thirteen percent of the strong-purpose group reported burning out very often or always, against 38 percent. Same economy. Same job market. Different reason to show up.

Now the money. Gallup's Q12 meta-analysis is the piece of research to put in front of anyone who tells you culture is a luxury. It pools 736 research studies across 183,806 business and work units, 3,354,784 employees, 347 organisations, 53 industries and 90 countries. Comparing top-quartile engagement units with bottom-quartile ones, the median differences are 23 percent in profitability, 18 percent in productivity measured by sales, 10 percent in customer loyalty and 22 percent in organisational citizenship. The avoided-cost side is bigger still: 78 percent in absenteeism, 63 percent in safety incidents, and 21 to 51 percent in turnover depending on the organisation. And 70 percent in wellbeing. Read that list again as a profit-and-loss statement, because that is what it is.

People over profit is not a discount you hand out. It is the cheaper way to operate — and the research has been saying so for twenty years.

$10 trillion Gallup's estimate of what low engagement costs the global economy — about 9% of global GDP — Gallup, State of the Global Workplace 2025

Here is the bridge between purpose and performance, and it is narrower than you think. Gallup's work on mission-driven workplaces found that only about four in ten U.S. employees strongly agree that the mission or purpose of their company makes them feel their job is important, and only 41 percent strongly agree they understand what their company stands for. Four out of ten people can explain why your company exists. Move that even slightly and the outcomes shift: Gallup reports a 10 percent improvement in employees' connection with the mission is associated with a 12.7 percent reduction in safety incidents, an 8.1 percent decrease in turnover and a 4.4 percent increase in profitability. You do not need a new mission statement. You need the one you have to be legible to the person answering your phone.

Which brings us to the cheque, and to the most misread number in this whole conversation. Americans gave 617.20 billion dollars to charity in 2025 — the first year ever above 600 billion. Individuals gave 64 percent of it. Corporations gave 7 percent, about 43.67 billion, and corporate giving grew 3.1 percent, the slimmest growth of any source and essentially flat once you adjust for inflation. Giving USA estimates corporate giving at roughly 1 percent of corporate pre-tax profits, below the four-decade average. Sit with that. For all the noise about corporate social responsibility, the cheque is the smallest lever a business owner holds. The payroll is bigger. The culture is bigger. The customer list is bigger. The room you can fill is bigger.

So here is the honest version. Impact is not what you do with the money after the business works. Impact is how the business works. The data says engaged teams earn more and cost less. The data says purpose cuts turnover intent by almost half and burnout by two thirds. The data says the mission only pays when people can actually name it. And the data says corporate cheques are about 1 percent of pre-tax profit, which means the cheque was never going to be the thing that changed anything. You already own the lever. It is payroll, it is the room you can fill, it is the list that answers when you write, and it is whether the person who works for you can tell a stranger what this is all for. Build the for-profit. Fund the mission from it. Then go home at a reasonable hour, because a founder who burns out helps nobody.

The Centre Spread · Facing Page

The Secret Weapon for Non-Profits

The hardest problem most non-profits have is not heart. It is distribution.

The mission is sound, the people are good, and almost nobody outside the room ever hears about it. Meanwhile the for-profit world has spent thirty years building the machinery that solves exactly that problem: production, logistics, data, follow-up, a reason for a stranger to pay attention. The machinery is not evil. It is just rarely pointed anywhere useful. So point it. What follows is three real organisations and how they stack, written plainly so you can copy the shape even if you never work with any of them. One builds the experience. One makes sure the market sees it. One is the mission the whole thing is for.

Start with the production layer. Activate 4 Impact sells one line: from big idea to red-carpet rollout. Under it, three words — Strategy, Production, Execution — and one more, worldwide. It is a R0cketShip Holdings and Crews Beyond Limits company. The build list is eight items long: retreats and offsites, events and experiences, business launch, growth and scale, teams and culture, scale and exit, real estate and development, and tourism and destination. The sectors it names are government and defence, federal agencies, non-profits and movements, and enterprise, hospitality and tourism. Krystalore Crews, MBA, is Executive Director — a 22-year Air Force veteran, retired Senior Master Sergeant, who built a 40-hour Pentagon curriculum and the computer-based training behind it used by roughly 180,000 people a year. Jeff Cline is the growth and technology architect.

That gap between those two seats is the actual insight, and it is worth saying slowly. A great room does not sell itself. Non-profits learn this the expensive way: they run a genuinely moving event, forty people leave changed, and nothing happens next because nobody captured who came, nobody followed up, and the market never found out it existed. The division of labour here is explicit. Activate creates the experience. Amplify uses growth technology so the market sees it. Two halves, deliberately separate, so neither one gets to be an excuse for the other. If your event was good and nothing came of it, you were missing the second half. If your marketing was loud and nothing came of it, you were missing the first.

Your secret weapon starts with a conversation.worldchangers.ai

The second half has a name and a front door. worldchangers.ai calls itself The Founder's Edge, a Krystalore and R0cketShip joint venture, and its headline is a sentence most founders have never been offered: People First. Tech-Backed. Change the world without needing a vacation from your life. Its three pillars are stated flatly — People First, growth that fits your life rather than the other way around; Tech-Backed, predictive data that shows who is in-market before competitors; and Change the World, scale your impact without burning out. The offer is a ladder rather than a package, and the top rung is simply called The Secret Weapon.

Then the third organisation, which is the reason the first two matter. Her Next Mission is the featured non-profit of this edition. It is a 501(c)(3) in formation — say that precisely, because it is not yet a determined 501(c)(3) and the difference is real for anyone considering a gift. It was founded by SMSgt (Ret.) Krystalore Crews, USAF. Its two public lines are From Service to Success and IT'S HER TURN. Its stated purpose is for female Veterans and first responders transitioning out of service, offering coaching, community and clarity for the next mission. The programmes named are one-to-one somatic and life coaching, group business and clarity bootcamps, multi-day immersive retreats, an annual summit and events, a podcast, and grants and donation programmes. The people it serves come out of military, fire, law enforcement and EMS.

Why this population, and why now. The Department of Veterans Affairs is blunt about the trend: more than two million women Veterans live in the United States today, and women are the fastest growing group in the Veteran population. In 2000 they were 4 percent of the Veteran population. By 2023 that was 11.3 percent, and it is projected to reach 18 percent by 2040. The median age of women Veterans was 52 in 2023 — these are not all young separations, they are women in the middle of a working life. About 930,000 were enrolled in VA health care in 2023, which is 45 percent of all women Veterans.

• 2.08M women Veterans in the U.S. in 2025 — 12% of all Veterans — source
• 4.6% vs 3.3% unemployment rate, female Veterans vs male Veterans, 2025 — source
• 11.3% to 18% women's share of the Veteran population, 2023 to projected 2040 — source
• 930,000 women Veterans enrolled in VA health care in 2023 — 45% of all women Veterans — source

The employment picture is where the gap gets measurable. The Bureau of Labor Statistics counted 2,077,000 women Veterans in 2025, 12 percent of all Veterans. The unemployment rate for female Veterans was 4.6 percent, against 3.3 percent for male Veterans — and 4.0 percent for nonveteran women. Read that triangle carefully, because it is the whole argument. Male Veterans do better than the women who served alongside them, and women Veterans do slightly worse than women who never served at all. Whatever service is doing for men on the way out, it is not doing for women. That is not a feelings problem. That is a transition-design problem, and transition design is a thing you can build.

One last thing, for the woman reading this who is six months out from her separation date and has not told anyone how lost she feels. Two million of you. The fastest-growing group in the Veteran population. Unemployed at a higher rate than the men you served beside, and at a higher rate than the women who never served at all. That is not your personal failure. It is a system that has not caught up to who serves in it, and it is being built right now by people who have been through it. From service to success is not a slogan on this page. It is the gap in the data, and somebody is finally working in it.

Safety, incivility and permission

The term comes from Amy Edmondson, and the definition in her 1999 paper in Administrative Science Quarterly is worth having exactly: team psychological safety is a shared belief held by members of a team that the team is safe for interpersonal risk taking. Read it twice. It is not about comfort, and it is not about being nice. It is a belief about consequences — specifically, whether you can ask the dumb question, admit the error, or disagree with the person who writes your review, without it costing you. Shared, because it is a property of the group rather than of your personality.

Her study covered 51 work teams in a manufacturing company, measuring antecedents, processes and outcomes. Psychological safety was associated with learning behaviour — asking questions, seeking feedback, discussing errors. Team efficacy, the belief that the team is capable, was not associated with learning behaviour once psychological safety was controlled for. And learning behaviour mediated the relationship between psychological safety and team performance as rated by managers. The antecedents mattered too: context support and team leader coaching shaped how safe the team felt. Leadership behaviour in, safety out, learning in the middle.

Now say the limitation out loud, because it is a field study and not an experiment. Fifty-one teams in one manufacturing company, measured at a point in time, with the outcome rated by managers who were not blind to anything. Mediation inferred from cross-sectional data is the kind of claim that has taken a beating across psychology in the twenty-five years since. The construct has held up extremely well and it has been replicated in a lot of settings. The specific causal pathway has not been nailed down by that design and Edmondson's paper does not claim otherwise. Treat the 1999 study as the thing that named a real phenomenon and showed it travels with learning and performance.

The randomised evidence comes from an unexpected place. In 2015, Arieh Riskin, Amir Erez, Trevor Foulk and colleagues published a randomised trial in Pediatrics. Twenty-four neonatal intensive care unit teams ran a simulated medical crisis. The teams were randomly assigned to one of two conditions. In both, a person presented as an expert on team reflexivity made comments during the simulation. In one condition the comments were mildly rude and had nothing to do with the team's performance. In the other they were neutral. Outcome assessment was blinded. That is a clean experiment on a question most people assume cannot be tested.

The teams exposed to mild rudeness performed worse. Diagnostic performance scored 2.6 against 3.2 for the control teams, with a p value of .005. Procedural performance scored 2.8 against 3.3, p equals .008. Rudeness explained 12 percent of the variance in both measures. And the mechanism was identified: a model including information-sharing and help-seeking as mediators explained 52.3 percent of the variance in diagnostic performance and 42.7 percent in procedural performance. Rudeness did not make the clinicians less competent. It made them stop talking to each other and stop asking for help. In a neonatal crisis.

Rudeness had adverse consequences on the diagnostic and procedural performance of NICU team members. Information-sharing mediated the adverse effect of rudeness on diagnostic performance, and help-seeking mediated its effect on procedural performance.Riskin et al., Pediatrics, 2015

Sit with the dose for a second. Mild comments. From an outsider who had no power over anyone. Unrelated to the work. During a simulation where nobody's actual patient was at risk and everyone knew it. And measurable degradation in clinical performance. If that is what a stranger's offhand remark does inside a controlled simulation, be serious about what a team lead who sighs at questions does across a fiscal year. This is the single strongest piece of evidence in this whole edition, and it is strong precisely because it is small, randomised, blinded and specific.

• 2.6 vs 3.2 Diagnostic performance, NICU teams exposed to mild rudeness versus neutral comments, p=.005 — source
• 12% Variance in both diagnostic and procedural performance explained by rudeness alone — source
• 71% Public-sector employees reporting some experience of workplace incivility in the previous five years — source
• 1 in 3 Uncivil acts instigated by the most powerful individuals in the organisation — source
• 10.4x How much more powerful than compensation toxic culture was in predicting attrition relative to industry — source
• 51 teams Sample in Edmondson's original psychological safety field study, 1999 — source

How common is it? Here the evidence is survey-based, and I will label every number as such. The benchmark study is Lilia Cortina, Vicki Magley, Jill Hunter Williams and Regina Langhout in the Journal of Occupational Health Psychology, 2001. They surveyed 1,180 public-sector employees. Seventy-one percent reported some experience of workplace incivility in the previous five years. As many as a third of the uncivil acts were instigated by the most powerful individuals in the organisation. Women endured incivility more frequently than men, though both genders showed similarly negative effects on job satisfaction, job withdrawal and career salience. Uncivil experiences were associated with greater psychological distress, while physical and psychological health indices were relatively unaffected.

Note what that last clause does. The authors reported a finding that cut against the strongest version of their own story — distress yes, broad health effects no — and they published it. That is the mark of a study worth citing. The limitations are real: one sector, cross-sectional, retrospective over five years, and now a quarter of a century old. It establishes that incivility is widespread and that it is disproportionately top-down. It does not establish a causal effect on health.

The practitioner data goes further and stands on thinner ground. Christine Porath and Christine Pearson reported in Harvard Business Review in 2013 that across fourteen years of polling, 98 percent of people had reported experiencing uncivil behaviour at work, and that by 2011 half said they were treated rudely at least once a week, up from a quarter in 1998. Those are not probability samples of the US workforce. They are polls of managers and employees the researchers had access to, often through executive programmes, and the comparison across years is not a fixed panel. The direction is useful. The precise percentages should not be quoted as national rates, and I am not going to quote them that way.

The same caution applies to the most-shared culture statistic of the last five years. In January 2022, Donald Sull, Charles Sull and Ben Zweig reported in MIT Sloan Management Review that a toxic corporate culture was 10.4 times more powerful than compensation in predicting a company's attrition rate relative to its industry. The analysis used 34 million online employee profiles and 1.4 million Glassdoor reviews from Culture 500 companies between April and September 2021. The three leading components of toxicity they identified were failure to promote diversity, equity and inclusion, workers feeling disrespected, and unethical behaviour. It is a large and clever dataset. It is also self-selected review text, measured at company level, predicting a company-level rate — not a randomised test and not a claim about why any individual quit.

There is one more honest thing to say about the mechanism, and it links this feature back to the first one. What rudeness destroyed in the NICU trial was information-sharing and help-seeking. What psychological safety produced in Edmondson's teams was learning behaviour — questions, feedback, error discussion. Those are the same thing described from two directions. The commodity being protected or destroyed is the willingness to say something that might make you look bad. That is not a feeling. It is an operational input, and it is the one input that no wellness app, no step challenge and no engagement dashboard can supply.

The question is not whether your people feel safe. The question is what happened the last time someone told you something you did not want to hear, because everyone in the room is still working from that answer. Culture is what you tolerate, and the evidence says the tolerating is mostly done at the top.

Advertorial · The featured non-profit

It Is Her Turn

Two million women veterans. The fastest-growing group in the veteran population. And a transition built for somebody else.

There is a particular silence that lands about six months after separation. The structure is gone — the rank, the unit, the thing you were unambiguously good at — and what replaces it is a job search in a language you do not speak, conducted by people who read your record and ask what you did in the real world. Men feel it too. The data says women feel it harder.

The Bureau of Labor Statistics counted 2,077,000 women veterans in 2025, twelve percent of all veterans. Their unemployment rate was 4.6 percent, against 3.3 percent for male veterans — and 4.0 percent for women who never served at all. Read that triangle slowly, because it is the whole argument. Men do better leaving service than the women who served beside them, and those women do slightly worse than women with no service at all. Whatever the transition is doing for men, it is not doing for women.

• 2.08M women veterans in the U.S., 2025 — 12% of all veterans — source
• 4.6% vs 3.3% unemployment, female veterans vs male veterans — source
• 11.3% to 18% women's share of the veteran population, 2023 to projected 2040 — source
• 52 median age of women veterans in 2023 — mid-career, not retiring — source

The Department of Veterans Affairs is blunt about the trend. Women are the fastest growing group in the veteran population: four percent in 2000, 11.3 percent by 2023, and a projected eighteen percent by 2040. Their median age in 2023 was 52, which means these are not young separations tidily absorbed by a first career — they are women in the middle of a working life, with mortgages and teenagers and a resume that stops making sense at the top.

From service to success.Her Next Mission

Her Next Mission was founded by SMSgt (Ret.) Krystalore Crews, USAF, who did twenty-two years and then did the transition herself. It serves female veterans and first responders — military, fire, law enforcement and EMS — with one-to-one somatic and life coaching, group business and clarity bootcamps, multi-day immersive retreats, an annual summit, a podcast, and grant and donation programmes. The stated mission is to walk every woman through the transition no one prepared her for, until she finds the next mission worth running toward.

If you run a business and have ever said you want to do something for veterans, here is the honest version of what helps. Not a flag on your website in November. A referral. A sponsorship of one seat at one retreat. An interview with a woman whose resume you cannot immediately parse. The gap in the data above is not a feelings problem — it is a hiring problem and a transition-design problem, and both of those are things a business owner can actually move.

Advertorial · Activate 4 Impact

Run the Company. Make the Impact. Same Hours.

Most owners treat the business and the giving as two budgets and two calendars. The ones who get anywhere treat them as one.

Here is the trap. You decide you want the company to stand for something, so you bolt an impact programme onto the side of it — a committee, a day of volunteering, a cheque in December. It runs on goodwill, goodwill is the first thing a hard quarter spends, and by the second year it is a line item nobody defends. The problem was never commitment. It was architecture. The impact was built as a cost rather than as part of what the business already does well.

The alternative is to point the machinery you already own at it. You have a payroll that clears, customers who answer when you write, a team that can run a room, and a reason for a stranger to pay attention. A non-profit has to raise its entire budget again every year from scratch. You do not. That asymmetry is the whole opportunity, and it is why the most effective impact work attached to a for-profit looks nothing like charity and everything like operations.

From big idea to red-carpet rollout.

Activate 4 Impact is built on that premise and staffed for it. Strategy, production, execution — worldwide. Eight kinds of work: retreats and offsites, events and experiences, business launch, growth and scale, teams and culture, scale and exit readiness, real estate and development, and tourism and destination activation. Four sectors: government and defence, federal agencies and the public sector, non-profits and movements, and enterprise, hospitality and tourism. It is a R0cketShip Holdings and Crews Beyond Limits company. Krystalore Crews, MBA, is Executive Director — twenty-two years in the U.S. Air Force, retired Senior Master Sergeant, who built a forty-hour Pentagon curriculum and the computer-based training behind it. Jeff Cline is the growth and technology architect.

• 200,000+ personnel trained worldwide — Activate 4 Impact's own figure — source
• $4M+ federal programs delivered — their own figure — source
• 2,000+ guests at a single production — their own figure — source
• 22 yrs of operational leadership — source

The division of labour is the part worth stealing even if you never hire anybody. Activate creates the experience — the retreat, the event, the launch, the programme. Amplify uses predictive demand, intent data, SEO, AEO and paid media so the whole market sees it and acts. Two halves, deliberately separate, so neither gets to be an excuse for the other. A moving event nobody hears about is a cost centre. A loud campaign for a mediocre room is worse, because it works once.

And the impact half is not theoretical here. The same team runs Her Next Mission for women leaving service, and worldchangers.ai for founders who want to grow without buying it with their health. One operator, one set of hours, a for-profit and a non-profit fed by the same machinery. That is the argument, and the page opposite is what it looks like when it works.