Stop Fixing Problems. Build the Ecosystem.

A customer acquisition problem appears. Marketing buys another tool.

Sales conversion falls. Management adds a new dashboard.

Customer retention weakens. A retention team is created.

Compliance becomes more complicated. Another vendor is hired.

Artificial intelligence arrives. Someone buys an artificial intelligence platform.

Six months later, the company has more technology, more vendors, more meetings, more data and more operating expense. The original problems may still be there.

This is a common organizational failure because each decision can make sense on its own.

The problem sits one level higher.

Businesses are systems of interdependent people, incentives, information, capital, customers, suppliers and institutions. When management repeatedly treats interconnected problems as isolated events, each local solution changes the environment around everything else.

The result is organizational patchwork.

There is another way to think about the problem: build an ecosystem.

That does not mean buying a platform or assembling a long list of partnerships. A functioning business ecosystem is an architecture in which different participants can contribute specialized capabilities while operating under sufficiently clear rules, incentives, interfaces and information flows that the pieces reinforce one another.

The distinction matters.

A patch solves today's problem.

An ecosystem changes the conditions that keep producing the problem.

Why humans build systems in the first place

Long before corporations, software platforms and supply chains existed, humans faced a basic economic constraint: individuals could rarely produce everything they needed alone.

Specialization created advantages. It also created dependence.

A good hunter could produce more meat than one household could immediately consume. Someone else might provide food later, childcare, political support, information, protection or access to another resource.

Once specialization appears, exchange becomes valuable. Once exchange becomes valuable, relationships matter. Once relationships become repeated, rules and reputations begin to matter.

John Q. Patton's research among households in Conambo in the Ecuadorian Amazon provides an unusually useful example.

Patton examined why hunters transferred meat to other households. A simple explanation might have been food redistribution. His findings were more complicated. Meat transfers were associated with reciprocity, kinship and political alliances. In other words, one resource moved through a network containing several kinds of value at once (Patton, 2005).

Meat could be food.

It could also reinforce reciprocity.

It could support relatives.

It could maintain an alliance.

It could contribute to status and reputation.

Patton explicitly cautioned against reducing a transfer to one motivation. A single act could serve several overlapping strategies depending on the ecological and political circumstances.

That observation translates surprisingly well into modern economic organization.

Money is rarely the only thing moving through a business relationship. Information moves. Trust moves. Customers move. Reputation moves. Risk moves. Expertise moves. Bargaining power moves.

A company that sees only the immediate transaction can miss the system producing its economic value.

Two competing explanations for how businesses should organize

There are at least two reasonable hypotheses.

Hypothesis 1: Specialization and modular purchasing are more efficient

Under this model, companies should solve problems individually.

Hire the best advertising agency for advertising.

Buy the best customer relationship management system for sales.

Use a separate vendor for compliance.

Find another specialist for analytics.

Replace components when better ones become available.

There is a strong economic argument for this structure. Specialization can reduce cost and improve expertise. Competition among suppliers can prevent complacency. Modularity can prevent companies from becoming dependent on one provider.

If this hypothesis dominates, fragmented systems should perform well when problems are largely independent and interfaces between them are simple.

Hypothesis 2: Interdependent problems require ecosystem coordination

This hypothesis starts somewhere else.

Marketing affects sales.

Sales practices affect compliance.

Product design affects conversion.

Conversion quality affects retention.

Retention affects customer lifetime value.

Customer lifetime value determines how much a company can rationally spend acquiring the next customer.

Data quality affects all of them.

Under these conditions, optimizing each component separately can produce a poorly performing whole.

The problem becomes one of coordination.

Research on business ecosystems by Michael Jacobides, Carmelo Cennamo and Annabelle Gawer describes ecosystems as arrangements in which independent organizations create value through complementary activities coordinated through common roles, standards and interfaces (Jacobides et al., 2018).

That definition is useful because it removes much of the corporate fashion surrounding the word ecosystem.

An ecosystem does not require one company to do everything.

Quite the opposite.

The architecture allows specialized participants to remain specialized while making their contributions compatible with the larger system.

The hidden cost of the patch

Reactive problem solving usually begins with a reasonable question:

What can fix this problem quickly?

The question becomes expensive when nobody asks the next one:

What caused this problem, and what else changes when we fix it?

Imagine a company struggling with customer acquisition costs.

Marketing buys cheaper leads.

On the marketing dashboard, cost per lead improves.

But cheaper leads convert poorly.

Sales responds by increasing call volume.

Customer complaints increase.

Compliance monitoring expands.

Agent productivity falls because employees spend more time contacting people unlikely to buy.

Management responds by buying another sales automation tool.

The company now has five separate problems.

In reality, it may still have one.

The original acquisition system rewarded marketing for generating inexpensive leads rather than economically valuable customers.

A local metric produced rational local behavior and an irrational system outcome.

This is one reason fragmented organizations accumulate software and vendors so easily. Each department can demonstrate the logic of its own purchase while nobody owns the economics of the complete customer journey.

The cost eventually appears somewhere: duplicated technology, integration work, reconciliation, additional headcount, compliance exposure, slower decisions or poor customer experience.

The balance sheet may call these separate expenses.

Operationally, many are coordination costs.

Ecosystems change the unit of optimization

An ecosystem approach changes the question from:

How do we make this department perform better?

to:

How do we improve the economics of the complete system?

Consider customer acquisition again.

A fragmented model might separately optimize:

traffic → leads → calls → sales → retention.

An ecosystem model treats those events as one connected economic sequence:

customer need → acquisition → qualification → recommendation → transaction → service → retention → learning → better future acquisition.

That final feedback loop is important.

A sale contains information.

Which customer converted?

Which advertisement produced the customer?

Which product matched the need?

Was the customer satisfied?

Did the customer remain?

Did the company make money?

Did the customer complain?

Without those signals flowing backward through the system, marketing continues buying customers using incomplete information.

The ecosystem learns only when outcomes return to the participants making upstream decisions.

This is one of the deepest economic advantages of connected systems: feedback improves allocation.

The ecosystem should be modular, not monolithic

There is an important objection.

Integrated systems can become bureaucratic.

They can create vendor lock-in. A dominant platform can extract excessive economic rents. Partners can become dependent on rules they do not control. Centralization can suppress experimentation.

Ecosystem research does not eliminate these problems. More recent work explicitly recognizes that platforms and ecosystems can generate their own functional and distributional failures after solving earlier coordination problems (Jacobides et al., 2024).

So the alternative to fragmentation should not be a giant vertically integrated machine.

The better architecture is usually modular integration.

Think of Lego rather than concrete.

Each component should have a defined purpose.

Interfaces should be standardized.

Data should be portable where practical.

Responsibilities should be explicit.

Participants should understand how value is created and how they are compensated.

Weak modules should be replaceable without destroying the whole architecture.

That is what makes ecosystems economically interesting. Independent organizations can coordinate without requiring a single hierarchy to control every activity.

Key terms

Fragmented architecture: Independent solutions optimized around individual problems, departments or metrics.

Ecosystem: A set of interdependent participants whose complementary capabilities combine to create a broader value proposition.

Modularity: Designing components so they can operate independently while connecting through defined interfaces.

Feedback loop: Information from downstream outcomes that returns upstream and improves future decisions.

Coordination cost: The time, money and organizational effort required to make separate actors and systems work together.

Why ecosystems can become economically stronger over time

A patch usually depreciates.

An effective ecosystem can learn.

Suppose a company connects acquisition data, customer behavior, sales outcomes, product performance, customer service and retention.

The first year of data improves the second year of decisions.

Better decisions improve customer selection.

Better customer selection improves economics.

Improved economics creates more room to invest in service, technology and distribution.

Those investments generate more information.

The system can begin compounding.

This does not happen automatically. Data without governance can produce confusion faster than insight. Partners with conflicting incentives can game one another. Network effects can also concentrate power rather than distribute value.

But when incentives and information are aligned, the economic properties change.

The company stops buying isolated capabilities and begins accumulating system capability.

That distinction may become increasingly important as artificial intelligence reduces the cost of individual tools.

If everyone can buy similar software, the software itself becomes less defensible.

What remains difficult to reproduce is the architecture surrounding it: proprietary workflows, customer relationships, trusted partners, operational data, distribution, regulatory knowledge, institutional memory and the feedback loops connecting them.

The competitive advantage moves from possessing tools toward organizing relationships among capabilities.

Evidence, interpretation and speculation

Evidence

Business strategy research increasingly treats ecosystems as a distinct organizational form for coordinating complementary but independently controlled activities (Jacobides et al., 2018). Research also suggests that ecosystem strategies can create value through expanding core businesses, adding complementary products and building end-to-end customer propositions.

The anthropological evidence gives us a much older version of the underlying coordination problem. Patton's Conambo research found that resource transfers were simultaneously embedded in reciprocity, kinship and political alliances rather than explained adequately by a single mechanism (Patton, 2005).

Interpretation

Modern companies face a structurally similar problem at a very different scale.

Economic value frequently emerges from interactions among specialized actors. Optimizing those actors independently works when dependencies are weak. As dependencies increase, coordination becomes increasingly valuable.

Ecosystem thinking therefore becomes most useful when customer outcomes depend on several organizations, technologies or functions working together.

Speculation

Artificial intelligence may accelerate this transition.

As specialized capabilities become cheaper to produce, competitive advantage may shift toward companies that can assemble, govern and continuously improve networks of those capabilities.

The scarce resource may become less about access to intelligence and more about coordination of intelligence.

That proposition remains to be tested.

Do not start by building an ecosystem

This may sound contradictory, but businesses should resist announcing an "ecosystem strategy" before understanding the economic problem.

Start with the customer or economic outcome.

Then work backward.

What does the customer actually need?

Which capabilities produce that outcome?

Where does information break?

Where do incentives conflict?

Which participant benefits while another absorbs the cost?

Where is the same work being repeated?

Which decisions require information trapped somewhere else?

Which components genuinely need integration, and which should remain independent?

Only then should architecture follow.

A useful operating sequence is:

Outcome → capabilities → dependencies → incentives → interfaces → data → feedback → governance.

Technology comes after those questions.

Otherwise the organization risks building a more sophisticated patch.

What would change my mind?

• Evidence that highly fragmented organizations consistently outperform coordinated architectures when customer journeys contain substantial cross-functional dependencies.
• Evidence that the coordination costs of ecosystems systematically exceed the value created by shared information and complementary capabilities.
• Evidence that modular architectures cannot meaningfully reduce ecosystem lock-in and concentration risk.
• Evidence that downstream customer and economic outcomes provide little useful information for improving upstream decisions.

Key takeaways

• Fragmented fixes can be individually rational while producing poor system-level outcomes.
• Ecosystems become valuable when economic problems are strongly interdependent.
• The objective is not vertical integration. Strong ecosystems preserve specialization while creating common interfaces, incentives and feedback loops.
• Customer outcomes should be the unit of optimization rather than departmental metrics.
• Information flowing backward through the system can make an ecosystem improve with use.
• Artificial intelligence may make individual tools easier to acquire, increasing the strategic value of the operating system connecting people, technology, data and partners.
• Ecosystems can create new dependencies and concentrations of power, so governance, portability, competition and consumer protection belong in the architecture from the beginning.

The economy has always been an ecosystem

Modern business language can make ecosystems sound new.

They are not.

Human economic life has always depended on networks of specialization, reciprocity, reputation, cooperation and exchange. What has changed is the scale and speed at which those relationships can now be coordinated.

A hunter transferring meat, a merchant extending credit, a manufacturer coordinating suppliers and a digital platform connecting thousands of independent producers inhabit radically different economies.

They share one constraint: value rarely exists in isolation.

It emerges through relationships.

Businesses that understand this can stop treating every problem as another hole requiring another patch. They can identify the recurring relationships underneath the problems and design an architecture in which information, incentives and capabilities reinforce one another.

That does not eliminate problems.

It creates a system capable of learning from them.

And over time, that may be far more valuable than getting better at fixing the same problems again.

---

References & further reading

• Jacobides, M. G., Cennamo, C., & Gawer, A. (2018). Towards a theory of ecosystems. Strategic Management Journal, 39(8), 2255–2276.
• Jacobides, M. G., Cennamo, C., & Gawer, A. (2024). Externalities and complementarities in platforms and ecosystems: From structural solutions to endogenous failures. Research Policy, 53(1).
• Jacobides, M. G. (2022). How to compete when industries digitize and collide: An ecosystem development framework. California Management Review, 64(3).
• Patton, J. Q. (2005). Meat sharing for coalitional support. Evolution and Human Behavior, 26(2), 137–157.
• Nowak, M. A., Tarnita, C. E., & Wilson, E. O. (2010). The evolution of eusociality. Nature, 466, 1057–1062.
• Foley, R. (1996). The adaptive legacy of human evolution: A search for the environment of evolutionary adaptedness. Evolutionary Anthropology, 4, 194–203.
• Ivanov, D., & Dolgui, A. (2022). Stress testing supply chains and creating viable ecosystems. Operations Management Research, 15, 475–486.

---

Editorial note: This article interprets business strategy research and evolutionary anthropology. Evidence, interpretation and speculation are labeled separately. The article reflects the author's analysis and should not be taken as specific business or legal advice.