How Do You Stay in Control When an Organization Becomes Too Large to Control Everything?
What the organization of living systems can teach us about steering, growth and regulation in large organizations
There is a rather pleasant problem to have: that of an organization that works.
It finds new resources. Activities increase. So do requests. A team that used to work in one city now works across several regions. An association becomes national. A company that used to make one product now makes ten. A humanitarian organization opens new field operations, recruits new skills, creates specialized departments. What could still be handled among a few people must now pass between professions, offices, countries and different responsibilities.
This development is generally sought after. It is even often regarded as one of the signs of success.
Yet something gradually changes in the very nature of the organization.
With ten people, the person in charge can still know almost every case.
With fifty, they generally know the people who know the cases.
With five hundred, they already depend on other people to learn what is happening, understand what is being reported to them, distinguish the important problem from the spectacular problem, decide, and then find out what became of the decision.
With several thousand people, the organization can end up generating a considerable share of its internal activity simply to allow its leadership to continue operating as it did when the organization was small.
Information flows upward, dashboards appear, meetings multiply.
Approvals are introduced, then levels of approval to make the first approvals more reliable. One person is tasked with coordinating two teams that were struggling to work together. Then another is needed to coordinate the coordinators. A procedure is created following a problem. It joins the procedures created following previous problems.
Taken separately, each of these choices may be rational.
The interesting phenomenon may lie elsewhere.
At what point do the mechanisms created to maintain control begin to reduce the actual capacity to control?
Growing is not simply becoming bigger
It is easy to think of growth as an increase in quantity.
An organization has one hundred people. It recruits fifty more. It has therefore become an organization of one hundred and fifty people.
Living systems do not grow quite like that.
A ten-ton animal is not a one-ton animal in which every cell has become ten times larger.
Growth produces differentiation. Some functions are strengthened, others appear. Certain structures specialize. Circulation networks must expand. The system must simultaneously maintain a boundary, receive matter, energy and information, process them, distribute what its different parts need, produce, store, eliminate, defend itself, repair itself, reproduce and continue to perceive what is happening around it.
In Les Idées du Vivant, I chose to look at human organizations in this way: not first through the organizational chart, but through the functions they must perform in order to continue to exist.
This shift is useful.
In a traditional organizational chart, we ask:
“Who is responsible for what?”
In a functional reading, we can also ask:
“What must continue to function, wherever it happens to sit in the organizational chart?”
The two questions do not necessarily produce the same map.
The function that enables an organization to perceive its environment is not contained within a single department. Sales receives information from customers. A field team sees a local transformation before headquarters does. Human resources sees certain signals emerge through recruitment or departures.
A scientist detects a technical development. A lawyer sees a regulatory risk approaching. Someone responsible for purchasing notices that a resource is becoming difficult to obtain.
All of them contribute, from different places, to the same major function of receiving information.
But what has been received still needs to be decoded.
Receiving is not understanding
This is a fairly obvious distinction in an organism, and curiously less obvious in organizations.
An eye receives light.
By itself, it does not “see” a truck approaching.
It transforms a signal that must then be processed, combined with other information and interpreted. An organization can also receive enormous amounts of information without understanding what is happening to it. It can even become extraordinarily effective at collecting information.
Dashboards are completed on time. Indicators are up to date. Reports move upward. Minutes exist. Everything is stored somewhere.
And yet an important change can remain invisible for several months because that change did not fit into any expected category.
This is one of the difficulties of growth: the more information an organization produces about itself, the less the existence of that information guarantees that it will be noticed.
Leadership can therefore receive more and see less.
The problem does not necessarily come from the quality of the people. It may come from the architecture of the system.
A very simple exercise could be carried out in many organizations.
Take a recurring report.
Then ask a single question:
What is the last important decision that this report actually changed?
If no one remembers, that does not automatically mean the report is useless. It may have legal, historical, accounting or scientific value.
But the question becomes interesting.
And the reverse is even more interesting:
what important information could appear somewhere in the organization today without any mechanism enabling leadership to understand that it is important?
Control is a function. It is not necessarily a place
When an organization is small, several functions can be combined in the same person.
The founder hears from a customer. They immediately understand the problem. They know what resource is available. They decide. They call the person who will carry out the action. They then check whether it worked.
Reception, decoding, decision-making, allocation of resources, regulation: a few minutes of conversation made all of this work.
When the organization grows, it is possible to try to preserve the same mechanism.
More and more situations are then escalated to the person who knows.
That person becomes very important.
They are often extraordinarily competent. They know the history of the organization, the people, old agreements, risks, exceptions to the rules and sometimes the reasons why those rules were written.
They can solve in ten minutes what would take others two days, and that is precisely how the system gradually learns to depend on them.
The more problems they solve, the more problems are brought to them. The more problems are brought to them, the more indispensable they become. And the more indispensable they become, the more rational it seems to continue escalating problems to them. It is a very effective feedback loop.
Until the day their effectiveness itself becomes a limit to growth.
The right question is probably not whether this person exercises “too much power”. That vocabulary quickly leads to a rather unproductive ideological debate.
A functional question is more useful:
how many essential functions has the organization gradually concentrated in this person?
Memory?
Decoding?
Arbitration?
Authorization?
Relationships with certain stakeholders?
Risk detection?
Transmission of culture?
Coherence between several activities?
If this person disappears for three months, it is not their titles that interest us.
It is the functions that stop.
Deciding, authorizing, being alerted, knowing
Confusion between these four situations probably produces a considerable share of bureaucracy in large organizations.
Leadership may need to know that a decision has been made.
That does not mean it needed to make it. It may want to be alerted when a situation exceeds a certain level of risk.
That does not mean it has to examine every situation below that threshold. It may need to authorize an irreversible or unusual commitment.
That does not mean it needs to design the solution itself. And some decisions do indeed need to remain within its remit.
It then becomes possible to retain a great deal of control over the system while reducing the number of decisions that reach the center.
This is less contradictory than it may seem.
A driver does not control their car by personally adjusting fuel injection several thousand times per minute. They have systems that regulate this operation and a few pieces of information that allow them to intervene when something leaves the normal range.
The human body takes this organization much further. A vast amount of regulation takes place without conscious decision-making. Temperature changes, blood vessels react, chemical concentrations change, cells locally correct disturbances. Several mechanisms intervene simultaneously, sometimes in opposite directions.
The brain is not absent; it is simply not responsible for doing everything.
And this allows it to process other things.
Escalating decisions may not be the right object
In many organizations, people try to define how much someone is allowed to spend.
1,000 euros here.
10,000 euros there.
100,000 euros higher up.
This is useful, but very partial.
Two decisions involving 10,000 euros can have radically different systemic meanings. One buys something the team purchases every month. The other may legally commit the organization for several years.
An almost cost-free decision can affect its reputation. A local technical modification may have major safety consequences.
The logic of living systems therefore leads us to shift the question again.
Which events should trigger a change in the level of regulation?
That is not exactly the same as “who has the right to decide?”.
A team can have a high degree of autonomy as long as it remains within certain limits. A piece of information can automatically trigger an alert. A particular risk may require an external opinion.
A completely new situation may fall outside established rules and be escalated immediately. The boundary becomes more intelligent. A biological membrane is not a wall.
It lets some things through, retains others, recognizes substances, maintains a difference between inside and outside.
An organization needs the same kind of function.
The more autonomous a unit is meant to be, the clearer its boundary must be.
What resources can it commit? What risks can it take? What belongs within its scope? What must be shared with others?
From what event onward must the higher level be informed?
Vague autonomy creates anxiety. Autonomy with precise boundaries can, on the contrary, increase leadership’s visibility.
What looks like duplication
Organizations have learned to seek efficiency by eliminating duplication.
The intuition is excellent as long as the duplication is genuinely redundant.
Living systems are much less economical from this point of view.
Two kidneys.
Two lungs.
Overlapping regulatory circuits.
Information received simultaneously from several places.
Different species sometimes performing similar functions within an ecosystem.
This apparent inefficiency has an interesting property: something can disappear without the function immediately disappearing with it.
This can be called useful redundancy.
Applied to an organization, it obviously does not mean recruiting two people for every position.
Once again, the question concerns functions.
Does a vital function depend on a single department? Can critical information only be detected by one team? Does an essential external relationship depend on one person? Is a core capability genuinely transferable?
When an entire function is confused with a single organ, the organization can appear extremely efficient until that organ disappears.
Robustness sometimes looks like an unnecessary cost as long as nothing happens.
That is true of almost every form of insurance.
Each department is not an isolated organ
The biological analogy becomes misleading if we decide that finance is the blood, leadership the brain, IT the nervous system, and so on.
A real organization is more interesting than that. The same department performs several functions. The same function is distributed across several departments. Leadership regulates, of course.
But an accounting procedure also regulates.
Peer review regulates.
A field team that refuses a technically impossible decision contributes to regulation. An IT system can prevent an unauthorized operation. A customer, member or beneficiary can trigger a correction.
In a sufficiently developed organization, the quality of regulation therefore depends less on the power of a central organ than on the quality of the relationships between all the functions involved in that regulation.
This in no way removes leadership. It changes its work.
It can devote less energy to handling every event itself and more to creating the conditions under which ordinary events will be handled correctly elsewhere.
That is a system-design function.
Growing like a cell?
There is another way to look at growth.
A cell reaches a certain size. It does not generally become a gigantic cell. It divides.
Human organizations sometimes experience something comparable.
Imagine an activity that has become very large.
One possibility is to continuously enlarge the same department. People are recruited. More managers are needed. The managers are grouped under another manager. Interactions increase.
Central functions must absorb the additional volume.
We call this growing, and it is indeed a form of growth.
Another possibility is sometimes to replicate a relatively complete unit.
A new team receives a mission, resources, skills and a sufficiently coherent decision-making space to solve a large proportion of the problems it will encounter itself.
The two units remain connected.
They can use shared resources.
They share a memory, certain standards, perhaps an identity, a technical infrastructure.
But their growth no longer requires every new person to interact with all the people already present.
This difference becomes enormous as size increases.
An organization that keeps adding connections ends up devoting a growing share of its energy to managing those connections. A modular organization can add capacity without increasing the number of interfaces to manage in the same proportions.
This is one of the questions proposed in the tool Growing Without Slowing Down, proposed by the Geneva Forum:
if your activity increased by 20%, by how much would the amount of coordination required increase?
The answer sometimes says more about the organization’s future capacity than its available budget.
The pyramid is not dead
It would be tempting to conclude this reasoning by deciding that hierarchical organizations are old and bad, while networks are modern and desirable.
Living systems do not teach us that lesson.
They use several forms at the same time.
Some functions must be extremely controlled.
A surgical operation does not necessarily benefit from holding a vote on where to make the incision.
A safety chain may require perfectly identified responsibilities.
An emergency situation may temporarily justify a strong concentration of decision-making.
In other situations, information is so distributed and the environment changes so quickly that a center simply cannot receive, understand and process everything that is happening.
A viable organization therefore probably knows how to use several geometries.
A mechanical and hierarchical part where repeatability and safety require it.
A biological and networked part when different units need to adapt locally while remaining coherent.
Sometimes something even more fractal, when small units emerge, recombine and disappear around problems that no one could have predicted.
The problem begins when a single architecture is applied everywhere because it feels reassuring.
The unique role of purpose
If not all decisions are escalated anymore, something must allow local decisions to remain compatible.
In an organism, this coherence is obviously not produced by sending a daily strategic memo to every cell.
Cells have common mechanisms, signals, constraints, a history embedded in the way they function.
In a human organization, part of this role is played by purpose.
The word “mission” is often used for communication.
It can serve a much greater purpose.
A mission becomes a genuine regulatory function when it enables decisions.
Two teams encounter two solutions that are equally legal, equally fundable and technically feasible.
One better serves the collective project.
Do they know which one?
If so, part of the arbitration has been distributed without losing coherence.
If not, someone will probably have to arbitrate above them.
An organization whose purpose is extremely clear can therefore allow much more autonomy.
It is not less controlled.
Part of the control has been embedded in the very criteria from which everyone makes decisions.
This phenomenon is familiar in organizations with a very strong culture.
The person who has just arrived asks for authorization.
The person who has deeply understood the project already knows what is compatible with it.
Of course, culture can also produce collective blind spots. Any sufficiently powerful mechanism can produce its own pathologies.
Hence the need for information coming from outside, diversity, contradiction and the capacity to question assumptions.
Leadership work changes with scale
In a small organization, leading can largely consist of making decisions.
This model works because the variety of problems remains compatible with the variety that one or a few people can absorb.
As the system grows, leadership progressively becomes something else.
Defining boundaries. Ensuring that critical information can circulate. Seeing that certain functions have become too dependent on one person. Knowing where local regulation is sufficient and where a central decision remains essential.
Observing deviations. Maintaining a sufficiently precise purpose. Deciding exceptionally and very quickly when the system moves outside its normal range.
The leader progressively stops being the person who handles a large share of events.
They become more the person who ensures that the organization knows how to handle events.
The difference is considerable.
And it explains a paradox.
Leadership can make fewer decisions while exercising far greater control over the whole.
Because boundaries are better understood, because exceptions become more visible, because a local decision leaves a trace, because a concerning signal triggers an alert instead of waiting for the next meeting, because functions no longer disappear with the people who carried them, because the top of the organization regains time to observe what the system does not yet know how to handle.
Five questions are sometimes enough to begin
There is no need to reorganize the company to check whether this issue exists.
Simply take a few real situations.
1. Which decision regularly reaches your level even though your intervention almost never changes the outcome?
If the answer is “many”, the escalation mechanism may be worth examining.
2. What important information could appear somewhere in the organization today without you being alerted?
The answer says more about actual control than the number of reports received.
3. Which essential function still depends on a person or team you would not know how to replace quickly?
It is a very simple snapshot of certain vulnerabilities.
4. In which situations does a team know exactly when it must stop deciding on its own and escalate the problem?
If no one knows the boundary, the organization will often compensate by multiplying approvals.
5. If activity increased by 50% next year, which part of your steering system would also have to increase by 50%?
That may be where the next limiting factor lies.
Take the “Growing Without Slowing Down” test.
A question I have been working on for a long time
In 2001, I published Les Idées du Vivant.
The starting point was fairly simple: billions of organisms and ecosystems function, develop, absorb disturbances, exchange matter, energy and information, specialize, cooperate, compete and sometimes disappear.
For our part, we spent enormous amounts of time designing human organizations while describing them almost exclusively through their power structures.
It seemed useful to try another angle.
Look at what living systems do.
Identify functions.
Look for their equivalents in a company, an association, an administration, a project.
Then observe what happens when one of these functions is absent, poorly connected to the others, oversized, monopolized by a single organ or unable to receive signals from its environment.
The proposal was not to turn companies into human bodies.
The analogy served as a framework for exploration.
In particular, it made it possible to look differently at information, boundaries, regulation, diversity, growth and the reproduction of organizations.
Twenty-five years later, some of these questions connect with my current research on collaborative governance and on ways of making these phenomena more observable and measurable.
One question has not changed much.
When the organization is small, the person who knows almost everything may indeed be the person who controls it best.
When the organization becomes immense, the proposition may eventually reverse.
The person who wants to continue knowing every problem, checking every arbitration and authorizing every decision may become, without intending to, one of the factors limiting the very system they are trying to protect.
The interesting question may therefore not be:
How far can I delegate?
It could be:
What do I actually need in order to retain control of the whole?
The two questions seem similar.
They produce very different organizations.
View the book on Bio-Inspired Management of Organizations Les Idées du Vivant.