A Flat Organization Is Not a Connected Operating Model

Every team delivered. The organization did not.

The round widget worked perfectly. So did the square receptacle. Together, they failed.

Each team had met its requirements. Each had passed its quality checks. Each could defend its decisions. The problem became visible only at the handoff, when two successful functional outcomes had to become one enterprise result.

I have seen versions of this failure at both ends of the organizational spectrum.

In a founder-led business, the founder becomes the invisible adapter who makes disconnected work fit. In an enterprise, escalations, committees, and layers of meetings become the adapter.

The structures look different, but the design failure is the same: the work has been distributed, while the connections have been left to chance.

The small team that works until it does not

In a founder-led business, informality can be a rational advantage. A five-person team does not need the same governance machinery as a global enterprise. Adding a policy, committee, or approval step to every activity would slow the organization without improving the outcome.

The problem begins when leaders confuse low complexity with no need for operating design.

The team may know how to complete the work, but the knowledge lives in conversations, private messages, and individual memory. The process succeeds because the same people perform it repeatedly. It has not yet been tested against turnover, growth, an unusual customer request, or a week when the founder is unavailable.

That creates a hidden form of founder dependence. The founder may no longer perform every task, but the founder still holds the context that allows the tasks to fit together.

When an exception appears, employees ask:

• Is this still within the promise we made to the customer?
• Can I change the sequence or deadline?
• Does this decision create more work for someone else?
• Who can approve a tradeoff between quality, cost, and speed?
• Is this a one-time accommodation or a new way of working?

If those answers exist only in the founder's head, the organization has delegated activity without delegating judgment.

That is not a scalable flat organization. It is a centralized decision model wearing a collaborative appearance.

The enterprise that breaks work into pieces

Enterprise organizations usually respond to complexity through specialization. That is not inherently wrong.

Specialization allows people to develop expertise, manage risk, improve quality, and perform work at scale. Acar and Tuncdogan (2026) make an important distinction in their analysis of organizational silos: separation can support innovation in some conditions, while it becomes harmful in others. The effect depends in part on how interdependent the work is and whether the groups are aligned around the same goal.

The leadership mistake is assuming that if every functional unit performs its assigned work, the enterprise outcome will assemble itself.

It often does not.

One team optimizes for accuracy. Another optimizes for cycle time. A third protects technical stability. A fourth manages the customer commitment. A fifth interprets the regulatory requirement. Each decision may be defensible inside the team's own boundaries.

Across the full system, those decisions can be incompatible.

The organization then spends enormous energy collapsing 50 functional decisions into one action. Meetings multiply. Escalations rise. Work is returned for clarification. Leaders discover late in the process that teams were solving different versions of the same problem.

The structure may look distributed, but the decisions still converge at the top because no one designed authority at the intersections.

The round widget and the square receptacle

The simplest version of this failure looks like a handoff.

One team is asked to design and build a widget. The team completes the work. The widget meets the requirements it was given. Its quality checks pass. Its leaders can reasonably say the team delivered its outcome.

The widget is then handed to the next team, which has built the receptacle that must receive it.

The receptacle is square. The widget is round.

Both components work independently. They do not work together.

By the time the mismatch is discovered, each team has invested time, defended design choices, and built downstream dependencies. The conversation is no longer a simple calibration. It is now a negotiation about rework, cost, accountability, and delay.

This is often described as a communication failure. Communication is part of it, but the deeper issue is operating-model design.

Before either team began building, the organization should have defined:

• The enterprise outcome both components needed to produce
• The interface where the two pieces would connect
• The requirements that could be decided locally
• The requirements that required a shared decision
• The point at which the design assumption would be tested
• The person accountable for the end-to-end result

That is what a learning organization does differently. It does not wait until the final handoff to discover whether the theory works. It calibrates on the design, tests the highest-risk assumptions while change is still inexpensive, and uses governance checkpoints to protect the enterprise vision.

The purpose of the checkpoint is not to ask whether everyone completed their tasks. It is to ask whether the pieces still fit.

Flat is not the problem; disconnected is

The current market is moving toward more horizontal operating structures. PwC's 2026 Digital Trends in Operations Survey of 767 operations and supply-chain leaders found that 94% of respondents with siloed or partially integrated structures expected to move toward a more horizontal, networked model. Only 41% reported operating with a collaborative, horizontal structure at the time of the survey.

The same survey found that 89% said their technology investments had not fully delivered the expected results, with integration complexity leading the reasons. Although the report is framed around digital operations, the more important leadership signal is structural: organizations do not receive enterprise value merely by improving isolated functions or installing more capable technology. The parts must operate as a system (PwC, 2026).

That is why I do not believe the answer is simply more hierarchy.

Adding management layers can create the appearance of control while leaving the underlying interfaces unresolved. The round widget may receive three additional approvals and still arrive at the square receptacle.

Nor is the answer endless collaboration. Requiring every team to participate in every decision destroys the benefits of specialization and creates consensus as a substitute for accountability.

The goal is connected autonomy.

Teams should own their outcomes. They should also understand how those outcomes fit into the work before them, the work after them, and the vision of the organization as a whole.

Research on team cognition helps explain why. DeChurch and Mesmer-Magnus (2010) synthesized 231 relationships across 65 independent studies and found that collective cognition, including shared mental models and knowledge about who holds which expertise, was meaningfully related to team processes and performance.

That does not mean every employee must think identically. It means the team needs a sufficiently shared understanding of the objective, roles, dependencies, and operating environment to coordinate its different expertise.

At enterprise scale, the same principle should extend across functional boundaries. Each team can own a different component while maintaining a common picture of what the components must accomplish together.

Governance belongs at the seams

Governance is often treated as a layer placed above the work. A committee reviews a status deck. Leaders approve a recommendation. A red, yellow, or green indicator summarizes months of design decisions.

That form of governance can arrive too late.

The most valuable governance occurs at the seams where one team's decision changes another team's ability to perform.

Those checkpoints should focus on questions such as:

• Has the enterprise outcome changed?
• Are teams still working from the same assumptions?
• Does a local decision alter another team's requirements?
• Which risks have become more or less material?
• What evidence would cause us to change direction?
• Is the decision still reversible, and at what cost?

These are not executive questions by default. Decision authority should sit as close to the relevant knowledge as the risk allows.

The employee closest to the work should be able to make a routine local decision. Cross-functional teams should jointly own choices that change a shared interface. Executives should intervene when the decision changes strategy, creates material enterprise risk, or requires a tradeoff that no individual function has the authority to make.

Clear decision rights prevent two opposite failures. They keep routine work from climbing the hierarchy, and they keep consequential cross-team decisions from being made inside one silo without the people who must absorb the impact.

A four-part test for a connected operating model

I would test a flat or networked organization through four connection points.

1. Shared direction

Every team should be able to explain the enterprise outcome in plain language.

This is more than knowing the mission statement. People should understand what the organization is trying to create for the customer, which strategic tradeoffs matter, and how success will be recognized.

If one team believes the priority is speed while another believes it is error prevention, the disagreement should be made explicit before it becomes competing designs.

2. Explicit interfaces

Teams should define what they receive, what they produce, and what the next part of the system must be able to do with the output.

For a critical handoff, the acceptance criteria should be designed by both sides. The team creating the widget cannot define success without the team responsible for using it.

An interface is not only a document or a technical specification. It is an agreement about dependencies, timing, quality, assumptions, and responsibility when conditions change.

3. Bounded decision authority

Employees need to know which decisions they own, which decisions require coordination, and which thresholds require escalation.

The goal is not to route more decisions upward. It is to prevent ambiguity about authority at the moment a decision crosses a functional boundary.

An end-to-end owner or value-stream leader can help preserve the full outcome, but that role should not become a new bottleneck. Its purpose is to resolve cross-system tradeoffs and maintain coherence, not to approve every local choice.

4. Safe learning loops

Teams need recurring opportunities to test assumptions, compare emerging work, and raise a mismatch without fear of repercussions.

Frazier and colleagues' (2017) meta-analysis drew on 136 independent samples representing more than 22,000 individuals and nearly 5,000 groups. The study connected psychological safety with meaningful work outcomes, including performance and organizational citizenship behavior, beyond related factors such as leader relationships and engagement.

In an operating model, psychological safety is not the absence of challenge or accountability. It is the ability to say, "The round widget will not fit the square receptacle," while there is still time to change the design.

A governance meeting cannot function as a learning loop if employees believe that surfacing a concern will be treated as resistance, incompetence, or disloyalty. Transparency requires leaders to reward early truth, especially when the truth complicates the plan.

The vision test

The most powerful organizations test whether their operating model is understood beyond the leadership team.

Choose people from different levels and functions and ask each person:

1. What outcome is the organization trying to create?
2. How does your work contribute to that outcome?
3. What must happen before your work begins?
4. Who must be able to use what you produce?
5. Which decisions can you make without approval?
6. Which decisions require another team because they change a shared interface?
7. Where can you challenge an assumption or report that the pieces no longer fit?

If employees can explain only their task, their target, or their functional scorecard, they know their page of the story. They do not yet understand the story the organization is trying to tell.

That gap matters because strategies are not executed by the org chart. They are executed through thousands of decisions made by people who must interpret how local work advances a larger outcome.

Leaders should not expect identical answers. Different roles will see the system from different angles. But the answers should be compatible. They should reveal a shared direction, visible dependencies, and a common understanding of where authority begins and ends.

A flatter organization needs stronger connective tissue

Flat organizations are often praised for speed, autonomy, and access. Those benefits are real when the operating model supports them.

Without connective tissue, a small business becomes dependent on the founder's memory. An enterprise becomes dependent on escalations and committees. In both cases, local teams may perform well while the organization struggles to produce a coherent end-to-end result.

The answer is not to choose between central control and functional independence.

It is to design the connections.

Define the vision clearly enough that people can use it. Make the interfaces between teams explicit. Place decision authority where the knowledge and risk meet. Create learning loops where employees can surface a mismatch before it becomes expensive.

Then test the operating model through the people who use it.

Every team member should be able to articulate more than their page of the story. They should understand how their page advances the whole story, what the next page requires, and what to do when the plot no longer makes sense.

Flat is an org-chart description. Connected is an operating capability.

References

Acar, O. A., & Tuncdogan, A. (2026, April 3). When silos hinder innovation, and when they can help. Harvard Business Review. https://hbr.org/2026/04/when-silos-hinder-innovation-and-when-they-can-help

DeChurch, L. A., & Mesmer-Magnus, J. R. (2010). The cognitive underpinnings of effective teamwork: A meta-analysis. Journal of Applied Psychology, 95(1), 32–53. https://doi.org/10.1037/a0017328

Frazier, M. L., Fainshmidt, S., Klinger, R. L., Pezeshkan, A., & Vracheva, V. (2017). Psychological safety: A meta-analytic review and extension. Personnel Psychology, 70(1), 113–165. https://doi.org/10.1111/peps.12183

PwC. (2026, April 23). PwC's 2026 Digital Trends in Operations Survey. https://www.pwc.com/us/en/services/consulting/supply-chain-operations/library/digital-trends-operations-survey.html