COMPLEXITY GROWS NON-LINEARLY
Scale does not simply enlarge an organisation. It creates new interfaces, dependencies, exceptions and information requirements. Complexity rises when those relationships grow faster than the structure designed to govern them.
Complexity is not the same thing as size.
A larger organisation can remain relatively simple if its units are modular, decision rights are clear and dependencies are limited. A smaller organisation can be highly complex if every product, client, market and executive decision depends on several others.
The structural variable is interdependence: how many other elements must be understood, coordinated or approved before a piece of work can move. Organisational information-processing theory has long argued that uncertainty and interdependence increase the amount of information that must be processed during execution. If information demand rises but organisational capacity does not, performance degrades through delay, rework, escalation or slack.
New nodes create relationships, not only volume.
Add a product and the organisation may add supplier, inventory, pricing, content, allocation, quality and channel decisions. Add a market and it may add legal, tax, logistics, localisation and commercial interfaces. Add a management layer and it may create new approval paths as well as new capacity.
A real organisation is not a complete network and does not require every node to connect with every other. The formula simply illustrates why adding nodes can increase potential coordination relationships faster than headcount itself.
What matters is therefore the density and criticality of dependencies. Research using design-structure matrices in product development shows that clustering tightly interdependent work and reducing unnecessary cross-team dependencies can reduce coordination complexity and cost.
When structure lags complexity, organisations compensate with human effort.
The first response is often informal coordination: more messages, more meetings, more senior involvement, more checking. This can work for a period because experienced people carry missing structure in their heads. But the model does not scale cleanly.
BCG's work on organisational complicatedness describes a similar response: companies confronted with external complexity add layers, procedures, interfaces, coordination bodies and approvals. The intervention can make the organisation more complicated without actually increasing its ability to deal with the underlying complexity.
The organisation must match information demand with capacity.
Galbraith's information-processing view offers a durable design principle. When uncertainty rises, an organisation has two broad choices: reduce the amount of information that needs to be processed, or increase its capacity to process information.
The key is fit. More process is not automatically more capacity. A forum adds capacity only if it brings the right evidence and authority together to close a real class of decisions.
Govern the interfaces that matter; remove the ones that do not.
A complexity response should start by identifying critical dependencies rather than redrawing the org chart. Which decisions cross functions? Which product families share industrial resources? Which markets genuinely require local variation? Which approvals protect material risk and which simply duplicate judgement?
Good architecture does not eliminate complexity. It decides where complexity is worth carrying. A luxury product business may deliberately accept difficult craft, material and supplier relationships because they create authority. It should be far less tolerant of complexity created by duplicated approvals, unmanaged variants or unclear ownership.
Track complexity through its consequences, not through headcount alone.
Useful signals include decision lead time, number of handoffs, percentage of decisions escalated beyond their intended level, frequency of decisions reopened, number of product or market exceptions, unique suppliers/components, meeting load around recurring decisions and the volume of unresolved cross-functional dependencies.
These are not universal KPIs and should not be collapsed into a decorative score. They are evidence for a structural diagnosis: where is information demand exceeding the capacity or authority of the current system?
The practical rule
When growth adds a new product, market, function or capital condition, ask not only “What capability have we added?” but also “What new interfaces and decisions have we created, and who now has the authority and information to carry them?”