Why Your Best Employees Keep Leaving: The Hidden Structural Causes

Half of all employees who quit do so because of their manager. Seventy-five percent of reasons for voluntary turnover lie in areas that managers can influence.¹ These figures from Gallup research have been stable for years. Despite all the investment in leadership programs, feedback culture, and employee development, companies continue to lose their best people.

The typical response: find better managers or train existing ones. But what if the problem isn’t with individual managers but with the structure itself?

The Statistics Behind the Resignation Wave

The “Boss Effect”—the impact of direct managers on resignation decisions—is well documented. Gallup research shows that 70 percent of team engagement variance is attributable to the manager. When managers are disengaged, so are their teams.¹

The problem is intensifying: In 2024, the global percentage of engaged employees fell from 23 to 21 percent—the steepest decline since the COVID-19 lockdowns. Manager engagement dropped from 30 to 27 percent, particularly sharply among young leaders under 35.¹

But all generations share a common experience: feeling unheard in hierarchical structures.

The question is no longer whether management is a retention problem. The question is: why do even well-intentioned improvements fail to reverse the trend?

What Research Actually Shows

Academic studies on organizational structures and employee retention provide surprising answers. A study of 95 employees in Switzerland and Germany compared holacratically organized companies with traditional hierarchies.² The results:

• Illegitimate tasks: In self-organized structures, employees perceived significantly fewer of their activities as pointless or below their qualifications (score 2.49 vs. 2.78).
• Appreciation: Perceived recognition was significantly higher (5.33 vs. 4.14 on a scale to 7).

These differences don’t arise from better managers but from structural changes in how work is organized.

Another study on self-organizing teams examined the relationship between autonomy and engagement.³ The result contradicts the intuitive assumption that more autonomy is always better: an excess of perceived autonomy—more than the employee ideally wants—correlated with lower engagement and reduced satisfaction.

Research shows: it’s not about giving employees as much freedom as possible. It’s about creating the right fit between structural requirements and individual needs.

The Hidden Structural Causes

If half of resignations relate to management, but better managers don’t solve the problem—where does the cause lie?

Research identifies several structural factors that individual managers cannot influence:

Decision speed: In traditional hierarchies, decisions pass through multiple approval layers. A manager can be as well-meaning as possible—if every idea needs three months of coordination, it frustrates the best employees.

Role clarity without role rigidity: Employees need clear responsibilities but also the opportunity to develop them. Traditional job descriptions often create the opposite: rigid frameworks that provide neither clarity about current priorities nor room for growth.

Distributed recognition: In hierarchical structures, recognition concentrates on what supervisors see. In distributed structures, appreciation becomes broader and more authentic because it comes from actual work partners.

A study on radical decentralization at a Canadian technology company with 445 employees identified four critical factors for successful structural change:⁴

1. Authentic leadership—the change must be genuinely intended
2. Efficient administration—clear processes must work
3. Role clarity—who does what must be understandable
4. Role accountability—it must be traceable who is responsible for what

Only when all four factors were met did decentralization lead to lower turnover intentions.

Why “Better Managers” Isn’t Enough

The focus on individual managers overlooks a crucial point: managers work within structures that limit their options.

A middle manager in a classic hierarchy cannot simply give their team more decision-making authority—the structure doesn’t allow it. They cannot distribute recognition beyond their own area of responsibility. They cannot flexibly adjust roles when the HR system prescribes rigid job descriptions.

Research shows: the decisive factor isn’t the structure itself but the authenticity of its implementation.⁴ When employees feel that decentralization is merely pretend—that responsibility is formally delegated but informally controlled—turnover actually increases compared to honest traditional structures.

In other words: an honest hierarchy is better than a dishonest self-organization. But authentically implemented distributed structures outperform both.

Not for Everyone: The Personality Component

An important finding that research emphasizes: not everyone thrives in self-organized structures. The Swiss/German study found that employees with low neuroticism and high agreeableness fit better in holacratic organizations.²

This has practical implications:

• For recruiting: When hiring for self-organized teams, personality traits should be considered.
• For implementation: Not all departments need to be organized the same way. A gradual approach that considers employee readiness is more promising.
• For honesty: Some employees prefer clear instructions. This isn’t a weakness but a legitimate preference.

Research explicitly warns against mandating self-organization company-wide: “Self-management across all organizations cannot be the solution if it is only appropriate for a subset of the workforce.”⁵

The Zappos Warning

No discussion of organizational structures is complete without the most famous case: Zappos. The company implemented Holacracy in 2014 and subsequently experienced significant turnover: by May 2015, 14 percent of the workforce had left; by January 2016, it was 18 percent—a total of 260 employees.⁶

This example is often cited as proof against self-organization. But research paints a more nuanced picture: the increase in turnover wasn’t proof of decentralization’s failure but of the importance of structure-culture fit.⁷

At Zappos, those who left were primarily:

• Managers who had tied their identity to the traditional hierarchy
• Employees who couldn’t cope with the ambiguity of the transition phase
• People who never wanted to work in a decentralized environment

Zappos itself noted that about half of departures weren’t because of Holacracy but because the severance offer (minimum of three months’ salary or one month per year of service) was economically attractive.⁶

This turnover was partly healthy—a self-selection of those who didn’t fit the new structure. The problem wasn’t decentralization itself but the abrupt, company-wide implementation without sufficient preparation for cultural implications.

What This Means for Employee Retention

Academic research on autonomy, structure, and turnover suggests:⁸

First: The solution lies not in better individual managers but in designing structures that foster autonomy, competence experience, and belonging—the three basic psychological needs according to self-determination theory.

Second: More autonomy isn’t automatically better. The key is the fit between structural requirements and individual preferences.

Third: Structural changes must be authentically implemented. Cosmetic decentralization worsens retention compared to honest traditional structures.

Fourth: A selective, gradual approach is more promising than company-wide mandates.

If you want to reduce turnover among your best employees, don’t start with manager training. Start with an honest analysis of your organizational structure—and with the question of whether the way decisions are made at your company fits the people you want to keep.

Research on self-organized structures shows that alternatives to classic hierarchy exist. But it also shows that these alternatives don’t succeed automatically. Success depends on authentic implementation, cultural fit, and respect for individual differences.

---

Research Methodology

This analysis is based on peer-reviewed academic studies on organizational structures and employee retention. The cited studies use quantitative (N=95 to N=445) and mixed-methods designs, conducted in Switzerland, Germany, Canada, and the USA. Gallup research reports and documented case studies were additionally used for contextualization.

Disclosure

SI Labs has practiced Holacracy as an organizational model for over 10 years. This experience informs our perspective but does not influence the presentation of research findings, which include both positive and critical findings.

---

Sources

Footnotes

1. Gallup. (2025). State of the Global Workplace Report. gallup.com/workplace/349484/state-of-the-global-workplace.aspx ↩ ↩² ↩³

2. Holacracy, a modern form of organizational governance predictors for person-organization-fit and job satisfaction. Frontiers in Psychology (2023). DOI: 10.3389/fpsyg.2022.1021545 ↩ ↩²

3. Autonomy and engagement in self-managing organizations: exploring the relations with job crafting, error orientation and person-environment fit. Frontiers in Psychology (2023). DOI: 10.3389/fpsyg.2023.1198196 ↩

4. Turnover and Recommendation Intentions in the Post-Implementation Period of Radical Decentralization. Journal of Self-Governance and Management Economics (2022). DOI: 10.21818/001c.37162 ↩ ↩²

5. Does a self-managed organization leave employees behind? A critical review of the current trend. Development and Learning in Organizations (2024). DOI: 10.1108/dlo-10-2024-0322 ↩

6. Zappos Holacracy transition data: TIME Magazine (2016), “Zappos’ Weird Management Style Is Costing It More Employees”; OpenView Partners, “What’s Really Going on at Zappos?” ↩ ↩²

7. Holacracy – the future of organizing? The case of Zappos. Human Resource Management International Digest (2018). DOI: 10.1108/hrmid-08-2018-0161 ↩

8. The Impact of Radical Self-Management: The Formation of Organizational Culture that Supports the Satisfaction of Basic Human Needs in Holacracy. Emerald Publishing (2025). DOI: 10.1108/978-1-83708-974-120251003 ↩