Why Structure Determines Strategy
One Takeaway: Organizational structure isn’t really about org charts or reporting lines. It’s about economic trade-offs between specialization and coordination. Understanding these trade-offs explains when to separate functions versus integrate them. It heps explain when hierarchy can be useful and when it fails. It also helps with how to build structures that let operators, refiners, and creators work at the same time without destroying each other.
Throughout this series we’ve explored why coordination becomes expensive (”The Cost of Working Together”), how knowledge distribution affects decisions (”What Headquarters Can’t See”), why measurement systems create systematic biases (”Why Metrics Can Kill Innovation”), and why a fundamental job of organizational leadership is designing rules that make cooperation rational rather than hoping for motivated compliance (”Bad Rules Beat Good People”). But all of these insights need organizational structure to actually work.
You can understand the idea that operators, refiners, and creators need different systems. But, if your structure forces them into uniform processes, that understanding doesn’t help.
You can recognize that some decisions need decentralization. But, if your structure centralizes authority, recognition doesn’t matter.
You can design thoughtful measurement systems and well-aligned incentives. But, if your structure makes coordination impossible, none of it matters.
Structure is where institutional design becomes real. It’s how the rules of the game actually get implemented. Get it wrong and success becomes near impossible no matter how good your strategy, people, or resources.
Two Companies, Same Strategy, Different Structures
Two SaaS companies pursued identical strategies: maintain a profitable core product while building a next-generation platform. Both understood they needed different systems for each. Both had smart leadership. Different structures produced radically different results.
Company A used a single product team with all work integrated within it. The same engineering team worked on both core product and the new platform. They used the same prioritization process for maintenance of the current system and innovation for the new product. They had a unified roadmap. They split resources. They had common success metrics.
The logic: “We’re one company with one strategy. Integration enables sharing best practices and maximizes resource efficiency.”
What actually happened: In Q1, the new platform got 40% of engineering resources as planned. By Q2, critical bugs in the core product pulled resources away “just temporarily.” A major customer threatened to cancel their contract unless they got specific core product features. Project management prioritized measurable core product work. New platform resources dropped to almost nothing and became an afterthought.
This pattern repeated for months. Important new platform work was repeatedly deprioritized for urgent core product needs. Engineering productivity dropped. The platform shipped eighteen months late. By then, a competitor had captured the market. Company A was eventually acquired at a disappointing valuation.
Company B split the work into separated divisions. The Core Product Team got 70% of resources and focused on optimization of the current system that was generating revenue and clients. They had dedicated engineering. They planned in quarters. They focused on profitability and customer satisfaction. The New Platform Team got 30% of resources and focused on innovation. They had their own dedicated engineering. Planned in annual sprints. And their metrics focused on technical milestones and customer validation.
Company B assigned someone to bridge the two divisions. Their role was to think across the boundary rather than within each side. They made sure core product insights informed platform design. They prevented the platform from rebuilding what already existed. They managed resource allocation. And they translated between different evaluation frameworks. This was harmonizer thinking in practice. They didn’t create a new department. They simply made a deliberate commitment to having someone focus on the connection between the two rather than the success of either one alone.
What actually happened: When the core product hit critical bugs, the core product team solved them without pulling resources. When a major customer requested features, the core team delivered without affecting the platform. The team member thinking like a harmonizer made sure customer insights informed new platform strategy without creating resource conflicts.
After eighteen months: the core product achieved profitability improvements. Meanwhile, the new platform began customer pilots on schedule. Engineering was productive because teams had focused work. The company launched its platform on time, created competitive differentiation, and was eventually acquired at a premium valuation.
The difference wasn’t strategy. Or talent. Or total resources. It was structure. Company A’s integrated structure created conflicts that favored optimization over innovation. Every resource decision became zero-sum. Urgent measurable work always won. Company B’s separated structure enabled both at the same time. Different evaluation systems meant they weren’t competing. Physical separation prevented resource conflicts.
The Fundamental Trade-off: Specialization vs. Coordination
Every org structure solves the same economic problem that Adam Smith identified centuries ago. Specialization creates value but requires coordination. Coordination can become more expensive as specialization increases.
From earlier in this series, we know operators need reliability and clear processes. Refiners need systematic improvement and analytical tools. Creators need exploration and failure tolerance. When these functions get brought together, conflicts can emerge in terms of priorities, resources, and evaluations.
Operators demand stability that can prevent creator experimentation. Creator uncertainty can threaten refiner’s goals to precise system improvement. Urgent needs from operators can consume creator resources. Measurable refiner and operator work dominates unmeasurable creator work. Separation lets each function optimize for its own requirements without interference.
The thing is, business problems don’t care about your org chart. Separation can create its own costs when problems span across different teams. Separate teams often end up solving similar problems independently. Knowledge developed in one team doesn’t reach others. Different teams make contradictory decisions. Opportunities that require cross-functional work die because nobody owns the coordination.
The org design question is always the same: do the benefits of specialization exceed the costs of coordination? This is Coase’s question from “The Cost of Working Together” applied to structure. Separate when specialization benefits exceed coordination costs. Integrate when coordination benefits exceed specialization costs.
When to Separate
Separate functions when their requirements are fundamentally incompatible. The strongest case for separation shows up when:
Time horizons conflict (optimization evaluated quarterly, innovation over years)
Risk profiles conflict (optimization minimizes failure, innovation requires it)
Resource needs conflict (optimization is predictable, innovation is lumpy and uncertain)
Success metrics conflict (optimization measured on efficiency, innovation on learning)
When these conflicts exist inside a single company, optimization usually gets prioritized above all else. Not because managers don’t value innovation. Because the rules of the combined system (shared metrics, unified resource allocation, common evaluation frameworks) make optimization individually rational and innovation individually costly.
This is the measurement bias from “Why Metrics Can Kill Innovation” and the design failure from “Bad Rules Beat Good People” put together. The system isn’t broken. It’s working exactly as designed. It’s just designed for one type of work.
Effective separation means dedicated teams, dedicated budgets, different evaluation frameworks, and different processes. But separation without coordination is just fragmentation. This is where harmonizer thinking applies. Thinking in this way bridges separated functions so the organization gets specialization benefits without paying the full cost of lost coordination.
When to Integrate
Integration makes sense when coordination benefits exceed specialization costs. This happens when:
Functions need constant communication. Like customer success and sales, where handoffs occur daily and success depends on a shared customer understanding.
When specialization costs are low. Like different product lines in similar markets that share capabilities and can be evaluated on similar metrics.
When consistency matters more than adaptation. Like brand and marketing, where fragmentation confuses customers and duplication wastes resources.
Even when you integrate, you still need to acknowledge different needs inside the shared structure. Create sub-teams with specialized focus. Use time separation—innovation sprints alternating with reliability sprints. Design evaluation systems that support both optimization and exploration. Protect time and budget for innovation even within an integrated structure. Integration works when you can manage conflicts without a systematic bias toward one type of work.
When Hierarchy Can Help and When It Fails
Hierarchy is the default structure in large businesses for a reason. It reduces real transaction costs. Authority replaces negotiation. A boss decides instead of parties bargaining. It creates clear accountability. It enables standardized processes. These benefits explain why every growing company reaches for hierarchy first.
But hierarchy fails predictably when knowledge is distributed. From “What Headquarters Can’t See,” we know that hierarchy centralizes decisions while valuable knowledge is often local. Information gets filtered climbing the chain. Context gets lost in reporting. Decisions are made without the knowledge needed to make them well. The larger the organization, the more severe this failure becomes.
Hierarchy can work when decisions depend overwhelmingly on systematic knowledge. These are times when consistency matters more than local adaptation. They also apply when speed of decision matters more than quality of local information. Hierarchy fails when decisions depend on local, tacit, time-sensitive knowledge. These are times when local adaptation creates more value than consistency, and when innovation requires the kind of risk-taking that hierarchical control tends to prevent.
Beyond Hierarchy
When hierarchy fails, organizations need different coordination mechanisms. Oliver Williamson, who extended Coase’s work, argued that the choice between governance mechanisms should match the economic properties of the activity being governed. Routine, well-specified work fits hierarchical control. But the most valuable cross-functional work is often too complex for hierarchy and too uncertain for market mechanisms. It needs something else.
That something else is what Elinor Ostrom’s work pointed toward in “Bad Rules Beat Good People.” Cross-functional opportunities inside organizations look a lot like the shared-resource problems she studied. No single department owns them. Success requires multi-party cooperation. Neither hierarchy nor internal markets can coordinate them on their own. The five rules we drew from her research are exactly what harmonizer thinking works to bring inside organizations. As a reminder these rules were
Make cooperation individually profitable
Let affected parties design their own processes
Match rules to context
Use graduated consequences
Back the system with legitimate authority
The shared budgets, joint metrics, and reputation systems we discussed in “The Cost of Working Together” are those principles made operational. In practice, this means most conflicts get resolved at the working level. They use pre-agreed frameworks rather than escalating everything up the chain. People thinking like harmonizers, who look across boundaries rather than up through them, make this possible by creating agreements before disputes become crises. But it only works if leadership supports these efforts rather than overriding them. When executives arbitrarily override cross-functional decisions, people quickly learn that the real authority is elsewhere. The coordination system collapses.
Structural Requirements for Each Function
Given everything this series has covered, here’s how structure can enable the three types of work.
Operators need decentralized authority with process standards. Operator work depends on local, time-sensitive knowledge. Centralizing operational decisions kills effectiveness because the knowledge doesn’t survive the reporting chain. By the time central decision-makers review operational issues, circumstances have already changed. Give operators authority at the point of customer interaction. Let them use local knowledge without escalation. Provide them with process standards that enable coordination, knowledge sharing across operators, and centralized resources for common needs. The operator principle: autonomy within boundaries.
Refiners need centralized analysis with distributed implementation. Refiner work benefits from systematic analysis across contexts. Data patterns that only appear in aggregate or optimization opportunities that need to compare performance across locations. Centralize analytical capabilities, data access, and knowledge management. But distribute implementation authority so local teams can adapt improvements to their context. Pure centralization loses local knowledge and creates resistance. Pure decentralization loses systematic analysis and creates duplication. The refiner principle: centralized learning, distributed execution.
Creators need protected separation with strategic alignment. Creator work needs protection from optimization pressure. Dedicated teams that don’t compete for operational resources. Different budget processes. Different evaluation criteria. But creators can’t be isolated completely or they lose connection to customer reality and organizational capabilities. Strategic alignment and coordination with the rest of the organization, maintained through harmonizer thinking, keeps creator work grounded without subjecting it to the measurement and resource dynamics that would kill it. The creator principle: separation with connection.
Aligning Incentives With Structure
Structure enables different types of work. But incentives determine what actually happens. If structure separates innovation and optimization but incentives reward only measurable short-term results, structure fails. As we discussed in “Bad Rules Beat Good People,” compensation is one lever among many. Decision rights, information flows, authority structures, and evaluation frameworks all shape behavior independently of pay. Structure has to align all of these, not just the compensation plan.
The most visible misalignment is incentive time horizons. Most incentive systems reward measurable outcomes on short time horizons. This works for operator and refiner work. It destroys creator work. If everyone is compensated the same way, everyone optimizes the same way. That means optimizing for what’s measurable now rather than what’s valuable later.
The solution is matching incentive time horizons to the economics of the work.
Operators create value through reliability and quality on short cycles. Quarterly bonuses tied to operational performance make sense here. They reward what operators actually produce.
Refiners create value through systematic improvement on medium cycles. Annual bonuses tied to learning and capability building match how their value materializes.
Creators create value through exploration and option creation on long cycles. Long-term equity that vests over multiple years aligns their incentives with the time horizon of their actual contribution.
On top of role-specific incentives, shared participation in overall company success, equity or profit sharing, keeps specialization from becoming selfishness. Everyone benefits when the whole organization succeeds. This creates natural pressure toward collaboration even when functions are structurally separated.
The same logic applies to cross-functional initiatives. When a project needs creator exploration, refiner optimization, and operator delivery, give it a dedicated budget and shared success metrics. All participating functions get rewarded based on the project’s outcome. When collaboration is profitable for everyone involved, it stands a better chance at happening naturally. When only one function benefits, it requires forcing. This predictabily creates resistance and produces poor results.
Looking Ahead: Creative Destruction Inside the Firm
Understanding organizational design as an economic problem explains how to build structures that enable operators, refiners, and creators to work together. But there’s a tension we’ve been circling throughout this series that structure alone can’t resolve.
The systems that make operators and refiners effective (standardized processes, clear metrics, hierarchical coordination) are the same systems that kill creator work.
The freedom that creators need (uncertainty tolerance, failure acceptance, long time horizons) threatens the reliability that operators and refiners depend on.
Structure can separate these functions, but the underlying tension between optimization and innovation remains.
Next, we’ll explore Schumpeter’s creative destruction as an internal organizational challenge. How do you build organizations that can destroy their own successful approaches when the market demands it? How do you keep the structures that enabled past success from preventing future adaptation? And what role does harmonizer thinking play in managing the most fundamental organizational tension: the need to follow through on what works and explore what might work next at the same time?
The Bottom Line
Organizational structure is about economic trade-offs between specialization and coordination. Every structural choice represents a decision about which costs to accept.
Separation creates specialization but requires coordination. Integration enables coordination but limits specialization. Hierarchy enables quick decisions but limits local knowledge use. Decentralization uses local knowledge but creates consistency challenges. You can’t eliminate these trade-offs. You can only choose which ones to accept deliberately.
The practical implications follow from the economics of the situation and the roles involved. Separate innovation from optimization when their requirements systematically conflict. In a combined structure, measurable urgent work tends to almost always beat unmeasurable important work. Bring functions together when coordination benefits exceed specialization costs. This is helpful when constant communication, shared knowledge, and consistency matter more than specialized focus. Match structure to the work. Hierarchy for decisions requiring systematic knowledge. Peer coordination for cross-functional challenges. Decentralized authority for work that depends on local knowledge.
Structure alone doesn’t determine success. But the wrong structure makes success near impossible, and more dependent on luck than we might like. The right structure can better enable operators, refiners, and creators to all succeed at the same time which is what long-term success actually requires.