In 2023, U.S. companies spent nearly $900 billion replacing employees who quit, according to the Work Institute. The spending bought almost nothing. Most of those companies were not failing to retain people. They were succeeding at producing the system they had built.
This story plays out in thousands of organizations every year. The conventional response, better pay, better perks, better managers, treats the symptom while leaving the cause entirely intact. The cause is the system itself.
What U.S. companies spent replacing employees who quit in 2023, according to the Work Institute. Replacing a single employee costs between 33% and 200% of their annual salary. Yet most companies treat turnover as a hiring problem rather than a design problem.
The Numbers Don’t Point Where You Think
According to Gallup’s 2024 retention data, engagement and culture account for 37% of departures, and wellbeing and work-life balance account for another 31%. Together they explain nearly 70% of why people leave. Pay and benefits, the thing most organizations reach for first, account for just 9%. More tellingly, nearly half of employees who leave report that their employer could have done something to keep them. The window for intervention was there. It just was not visible because leaders were looking in the wrong place.
The standard narrative is that people leave bad managers. There is truth in that. Poor leadership ranks among the top three reasons employees quit. But the manager is not the cause. The manager is the output of the system that selected them, promoted them, and evaluated them on metrics that have nothing to do with developing the people around them.
Individuals are the inputs. The system is what determines the output.
What Structure Actually Does
Every organization has a formal structure: org charts, job descriptions, performance review cycles, compensation bands. The more powerful structure is the informal one. What actually gets rewarded. What gets ignored. What behavior the system makes structurally inevitable. When a company says it values work-life balance but its promotion criteria reward 60-hour weeks, the informal structure has spoken. When it says it values employee development but managers are evaluated only on output, not retention, the architecture has told the truth that the culture statement concealed.
Research confirms this. A 2024 survey found that 56% of employees believe their manager was promoted too quickly, and more than half believe their manager has not received appropriate training. This is not a management quality problem. It is a promotion criteria problem. The system is selecting for technical skill or individual output and then placing those people into roles that require an entirely different capability: the ability to develop others. The container was built wrong.
Design Failure
A Fortune 500 retailer with 24.9% annual turnover, among the highest of any sector, runs annual engagement surveys and quarterly one-on-ones. Managers are trained in feedback frameworks. Turnover has not moved in three years. The structural problem: store managers are evaluated on sales per square foot and shrink rates. Retention is not a metric. The system produces exactly what it measures.
Design Success
Companies with comprehensive retention strategies, those that build retention metrics into manager evaluations, create explicit career pathways, and tie compensation to team stability, achieve 87% higher retention rates and 67% lower recruiting costs, according to SHRM. The difference is not culture. It is architecture.
Distribution: Who Gets What
One of the most consistent findings in retention research is that high performers leave at disproportionate rates, not because they are unhappy with pay, but because they are unhappy with access. Access to interesting work, to visible projects, to sponsors who will advocate for their advancement. In most organizations, that access is distributed informally, through proximity to power and through networks that correlate heavily with demographics, tenure, and personality type. The quiet performer doing excellent work in a remote office loses to the louder performer who has lunch with the division head.
When organizations say “we lose our best people,” they are often describing a distribution problem masquerading as a retention problem. The talent did not leave because they were not paid enough. They left because the system did not give them what they came for. And the system’s distribution of visibility, sponsorship, and opportunity was invisible. Never named. Never measured. Never redesigned.
Adaptation: Why the Fixes Don’t Stick
Companies that respond to turnover with raises and surveys often find the system right back where it started six months later. This pattern, intervention followed by temporary stabilization followed by return to baseline, is one of the clearest signs of adaptive failure. The system absorbed the change and returned to its prior state because the architecture beneath the change had not moved. You cannot raise your way out of a system designed for departure.
Across U.S. industries, the Work Institute reports that 38% of employees quit within their first year, and 40% of those leave within the first 90 days. The data is not industry-segmented in any public source I could find, which itself is telling. We measure who is leaving. We do not measure what they were leaving from. Organizations that lose people this quickly do not have a talent problem. They have a structural onboarding problem. The architecture did not integrate new people into the system. It exposed them to it and let them draw their own conclusions. Research from SHRM finds that structured onboarding raises retention by 50% and productivity by 62%. Those gains are not produced by better culture. They are produced by better design.
The lowest-turnover sectors in the United States, government and financial services, do not have better people or better managers. They have systems that hold the relationship in place. Some of that holding is genuinely structural in a positive sense: clearer role definitions, predictable career trajectories, longer tenure norms, deep institutional knowledge. Some of it is structural in a less flattering sense: pension cliffs, non-transferable certifications, deferred compensation, and golden handcuffs that make leaving expensive even when staying is unsatisfying. Both are architecture. The architecture holds either way. The question every leader should ask is which version of the architecture they are building, the one that earns the relationship or the one that traps it.
The Design Question
“If you mapped every formal and informal incentive in your organization, what behavior would a rational person conclude the system was built to reward, and is retention one of them?”
The next time a high performer hands in their resignation, resist the instinct to ask what you could have done differently for that person. Ask instead what the system produces when it runs as designed. If the answer is turnover, then better pay and better coaching are rearranging furniture in a burning building. The building needs to be redesigned. Until it is, the resignations will keep coming, in exactly the volume the system was designed to deliver.
Is this a people problem or a design problem?
Describe your organization’s retention challenge and get a full Structure, Distribution, and Adaptation diagnosis specific to your situation.
Analyze Your System