Your Org Chart is fine. Your organisation isn't.
Lessons from Haier's RenDanHeYi model on why decentralised structures fail, even when the design is right
For founders and leaders building organisations without a traditional hierarchy
You restructured the organisation. You pushed decisions down to the teams closest to the customer. You removed management layers, redesigned incentives, and gave people real accountability for their own results.
And it is still not working the way it should.
Decisions meant to happen within teams still get filtered up. Accountability that looks clear on paper is diffuse in practice. Two teams with identical structures perform completely differently, and no one can explain why. The people who were supposed to step into ownership are still waiting to be told what to do.
I once watched a leadership team that had formally removed all titles. On paper, the organisation was flat. But every decision still paused when the former COO spoke. No one had authority over him anymore. Yet everyone waited for his reaction before committing.
Every organisation runs on two systems:
The formal system—structure, incentives, governance.
The informal system—power, loyalty, habits, history.
Structural redesign changes the first. But decentralisation only works if the second evolves too.
This is not a structural problem. The structure is right. It is a behavioural problem, and it will not be solved by refining the org chart.
The most advanced real-world model for this kind of organisational design is Haier's RenDanHeYi system. Haier is a $40 billion company that operates without a traditional hierarchy across 80,000 employees. If there is a ceiling to what structural redesign alone can achieve, Haier is close to it. And even there, the practitioners working within the model consistently identify the same gap: the hardest part is not the structure. It is the behaviour that the structure cannot reach.
What Haier built and why it matters
Haier grew from a near-bankrupt refrigerator factory in the 1980s into the world's largest home appliance manufacturer, not by scaling bureaucracy but by systematically dismantling it.
The RenDanHeYi model, developed over three decades by CEO Zhang Ruimin, organises the company as an ecosystem of thousands of self-managing micro-enterprises (MEs). Each ME controls its own hiring, compensation, and profit distribution. Each is directly accountable to customers, not to their managers. Support functions like HR, finance, and legal are not departments but independent MEs that other teams pay for on a market basis. If an ME consistently underperforms, it dissolves. Employees are not fired; they move to other projects or start new MEs. When several MEs need to collaborate on a larger opportunity, they form a temporary alliance, agree on profit-sharing in advance, and dissolve the alliance when the project is complete.
The model has been applied beyond China. The acquisition and transformation of GE Appliances in the United States is the clearest evidence that RenDanHeYi is not culturally specific.
For any founder trying to build an organisation that scales without becoming bureaucratic, this is the most important case study available. But it comes with a condition that its advocates do not always make explicit: Haier's own practitioners report that the transition to the model takes years, and the obstacles are almost never structural. They are behavioural.
As one Haier employee described it: "Unlearning old habits. Most people are trained to follow orders. Here, you must take ownership and act like an entrepreneur."
I address the disconnect between having a well-designed organisational structure and the far more difficult work of building the behaviours, habits, and relationships that actually make it function as intended. My focus is on what gets in the way of decentralised models working in real life, and how to recognise and work through those challenges as a founder or leader.
What RenDanHeYi actually solves and what it can’t solve
First, what does the model actually solve?
RenDanHeYi solves the organisational design problem. It answers the question: how do you structure a large company so that it retains the speed, accountability, and customer focus of a small one? The answer is to eliminate the layers that slow things down and replace them with small, autonomous units that operate like businesses within a business.
It solves the incentive alignment problem. Because ME teams share directly in their profits, individual and organisational interests are structurally aligned. There is no need for complex performance management systems; the commercial result of the ME does that work.
And it addresses the innovation problem. By running multiple MEs on similar opportunities and letting the best solutions win, Haier builds a competitive internal market that enables faster iteration than any centralised R&D function could.
These are genuine structural achievements. The model operates at an organisational design level that most companies never reach.
What it does not solve is the informal system that sits underneath the formal one: the established patterns of deference that persist regardless of what the org chart says, the founding team dynamics that set the performance ceiling for every team below them, and the accumulated history of the organisation that continues to shape behaviour long after the structure has changed.
Where it actually goes wrong
The following four problems are not theoretical. They are the patterns that appear most consistently in organisations that have restructured toward decentralisation and are not getting the results the structure should produce. The first two are context-setters. The third and fourth are where the real leverage is.
1. History still runs the organisation
When Haier restructured in 2012, it gave 12,000 managers a choice: leave or join the new structure. Many stayed. They did not leave their previous identity, their established relationships, or their informal sense of who carries weight and who does not. They brought all of that into the new structure.
In any organisation that has gone through significant change, there are people, teams, and periods that have been written out of the official narrative. A product line that failed. A co-founder who left badly. A business unit that was quietly shut down after a strategic error that no one discusses.
These unacknowledged elements do not disappear. They continue to influence decision-making and team dynamics in ways that are hard to trace but easy to observe: in the conversations that always stall at the same point, in the teams that consistently underperform without a clear commercial explanation, in the decisions that never seem to get made despite the authority being formally delegated.
In practice: Before restructuring, ask which parts of the organisation's history have never been explicitly acknowledged. A failed pivot. A difficult departure. A period of poor performance that the founding team would rather not revisit. These are not HR issues. They are operational drag. Naming them, even briefly and factually, reduces their pull on current behaviour.
2. Informal hierarchy replaces formal hierarchy
Remove the org chart, and something interesting happens. Authority doesn’t go away, but moves.
In every organisation, there is a sequence of entry: who was there first, who brought in the first significant client, who built the first version of the product. That sequence carries informal weight that persists long after formal titles have been removed.
When that informal hierarchy is not acknowledged, MEs have formal autonomy but real dependence. Decisions appear to be made by the team but are actually being filtered through the person whose informal status has not been recognised. Speed does not improve. Accountability becomes diffuse. The ME structure is in place, but the old hierarchy is operating underneath it.
In practice: Ask your team a direct question: when the formal structure is unclear, whose judgment do you actually defer to? The honest answer to that question maps the real hierarchy in your organisation. Those informal authorities need to be either formally recognised or actively transitioned out of the decision flow. Otherwise, the new structure papers over the existing power dynamic rather than replacing it.
3. The founding team sets the ceiling
This is where the greatest leverage lies.
RenDanHeYi was designed and built by one person, Zhang Ruimin, over three decades. The model reflects his thinking, his risk tolerance, and his beliefs about accountability and motivation. That is not a weakness; every organisation is shaped by its founders. But it is a variable the framework does not account for when applied elsewhere, because people implementing it in a different organisation bring their own patterns with them.
What transfers when a founding team implements RenDanHeYi is not just the structure. It is also the behavioural patterns of the people at the top. How the founding team handles disagreement. Whether real accountability is demanded or quietly softened when it becomes uncomfortable. Whether ME leaders who underperform are actually shut down or informally protected. Whether the profit-sharing model remains genuinely transparent or gets adjusted when it produces outcomes no one wants to discuss.
The practical observation across organisations attempting decentralised structures is consistent: the model performs to the level of behavioural clarity at the top. A founding team with unresolved tensions, unclear decision rights between themselves, or established patterns of avoiding accountability will replicate those patterns in every ME they create. The structure distributes those patterns downward. Everywhere.
In practice: Three diagnostic questions that surface the real dynamics in a founding team. First: when the founding team last genuinely disagreed on a significant decision, how was it resolved, and does everyone in the team describe that resolution the same way? Divergent answers identify where informal power is operating without acknowledgement. Second: Which decisions in the last six months took longer than they should have, and what was actually blocking them? The honest answer almost always points to an unresolved question about authority between founders. Third: if you applied the same accountability standard to your own work that you apply to your MEs, what would change? Founding teams that cannot answer that question cleanly are not yet ready to hold their ME leaders to genuine accountability.
4. Founders confuse architect and operator roles
The last area where significant leverage exists is role clarity at the founding team level.
Because RenDanHeYi removes traditional management layers, the founding team is simultaneously responsible for the design and performance of the system and, in many cases, for the direct operation of one or more MEs within it. Those three roles have different and often conflicting requirements.
The architect's role requires distance: the ability to evaluate the system objectively, to shut down underperforming MEs, and to change incentive structures when they produce the wrong outcomes. The operator role requires proximity: deep involvement in the commercial reality of a specific ME.
When founders conflate these roles, honest evaluation disappears. Underperforming MEs involving the founder or close associates are quietly extended. The willingness to shut down what isn’t working? It stops functioning where it’s needed most. And when the founding team’s MEs aren’t held to the standard, credibility erodes—fast.
In practice: Map every significant decision in the last quarter against the role it came from. Was it made from the architect position, evaluating the system's health, or from the operator position, protecting a specific unit's outcomes? Founders who find that the majority of their consequential decisions came from the operator role are functioning as participants in the system rather than as its designers. The structural fix is external accountability: a board, an advisory group, or a formal peer mechanism that evaluates the founding team's own units by the same criteria applied to everyone else.
How to Sequence the Work
The most common question founders ask once they recognise these patterns is practical: do I address the behavioural issues first, or implement the structure first?
The answer is neither strictly first. But the sequence matters, and the most common mistake is to implement the structure and then wait for behavioural problems to surface. By that point, the informal hierarchy has embedded itself in the new structure, the founding team's unresolved dynamics have set the performance ceiling for every ME below them, and role confusion has created accountability gaps that are difficult to unwind.
A more effective sequence works as follows.
Before restructuring, run the three diagnostics above. Map who the organisation actually defers to. Surface where the founding team's decisions are being avoided or informally protected. Identify the parts of the organisation's history that have never been acknowledged. This does not need to be a long process. It needs to be honest.
In parallel with restructuring, treat founding team dynamics as an ongoing operational discipline rather than a one-time exercise. The patterns that are invisible to ten people become organizationally significant at fifty and structurally determinative at two hundred. The ME structure will perform to the ceiling set by the people who designed it, and that ceiling needs to be raised continuously as the organisation grows.
After restructuring, build external accountability into the architecture from the start. Do not wait until the dissolution mechanism has failed twice before acknowledging that the founding team's own MEs are being held to a different standard. Integrate the external evaluation mechanism into the design before the first uncomfortable test.
The underlying principle is simple: whatever is unresolved in the informal system will reassert itself inside the new structure. Map it before you formalise it.
What happens after the Founder?
There is one question that the Corporate Rebels' research on Haier, some of the most thorough available on the model, does not yet fully address. Not because it is being avoided, but because it has not yet fully played out.
RenDanHeYi was designed, built, and refined over three decades by one person. The model carries Zhang Ruimin's intellectual fingerprint at every level. His convictions about motivation, the relationship between individual accountability and collective performance, and the customer as the only legitimate source of authority are not incidental to the model. They are constitutive of it.
The question is what happens when he is no longer the architect.
This is not a criticism of Haier. It is the most important stress test any organisational system can face: does it hold when the person who built it steps back? Can the principles Zhang Ruimin embedded survive as genuine organisational behaviour, or are they currently being held in place partly by his personal authority and the deference that three decades of demonstrated results command?
The answer will determine whether RenDanHeYi is genuinely self-sustaining or founder-dependent. That distinction matters enormously for any founder considering the model as a template, because it points to the work that structural design alone cannot do.
A self-sustaining system is one in which the behavioural patterns that make it work have been deeply absorbed into the culture, the incentives, and the daily operating practices, so that they persist without the founder's presence. A founder-dependent system is one in which the patterns are real and functional, but a significant portion of their force derives from the founder's personal authority rather than from the system itself.
Most organisations that claim to be the former are actually the latter. Recognising which one you are building is not a reason for discouragement. It is the precondition for building something that genuinely outlasts you.
The most useful question a founder can ask while implementing a decentralised structure is this: if I stepped back tomorrow, which of these patterns would hold and which would collapse? The patterns that would collapse are not evidence of failure. They are the roadmap for the remaining work.
Where structural solutions end, and the human challenges begin
RenDanHeYi is the most advanced real-world model for scaling without bureaucracy. The evidence from Haier and from GE Appliances confirms that it produces results at a scale and in a cultural context that few other frameworks have matched.
But the transition challenges that practitioners consistently report are not structural. They are behavioural. And behavioural problems require a different kind of intervention than organisational design can provide.
The organisations that get the most from RenDanHeYi treat the structural redesign and the behavioural pattern work as parallel workstreams, not sequential ones. They map the informal system before formalising the new one. They treat founding team dynamics as an operational priority, not a cultural afterthought. And they build external accountability into the architecture before the first test arrives, not after.
Structure redesigns the organisation chart. The behavioural pattern work redesigns the people and dynamics that fill it.
Both are necessary. Structures scale organisations. Behaviour determines whether they survive.
Based on the articles of Corporate Rebels
Wouter Gheysen is a systemic founder coach and transformation facilitator at Systemic Rebels. He works with founding teams and leadership groups at the intersection of organisational design and systemic patterns, where the formal structure is in place but the organisation is not yet performing to its design.