Why Systems Often Matter More Than Individual Events

The event is often only the visible surface

A familiar model in systems thinking compares a system to an iceberg. Visible events form only the tip. Beneath them lie recurring patterns, system structures and, deeper still, assumptions or mental models that help shape those structures.^[1]

Consider a sales organisation that misses its quarterly target. At event level, it is easy to identify specific explanations: a lost major account, a weak month, or a team that closed too few deals.

Across several quarters, however, a different picture may emerge. Perhaps the same pattern keeps returning: a slow start to the month, frantic closing activity at quarter-end and growing pressure to discount.

At that point, the individual weak month becomes less interesting than the question of why the same pattern keeps appearing.

The compensation system may reward short-term revenue far more strongly than margin or customer retention. The forecasting process may add further pressure towards the end of the quarter. Under such conditions, aggressive closing is no longer an inexplicable exception. It is a plausible response to the structure.

The event remains real. Its meaning changes once it becomes part of a pattern.

1. Patterns often contain more information than exceptions

A spectacular event can attract enormous attention while revealing relatively little about the system beneath it. A pattern is usually less dramatic, but often more valuable analytically.

If a production line fails once, a defective component may be a sufficient explanation. If it repeatedly fails under similar conditions, different questions become important: maintenance intervals, capacity planning, spare-parts availability or the way minor disturbances propagate through the production system.

The same logic applies inside organisations.

One resignation may be personal. Persistently high turnover in the same department points more strongly towards leadership, workload, compensation, career paths or recruitment practices.

The individual case still matters. The pattern tells us more about how the organisation works.

Systems thinking therefore often begins with a simple change of timescale. The question is no longer only “What happened?” but also “What keeps happening here?”

Extending the time horizon changes the diagnosis. What initially looks like an isolated incident may turn out to be the latest expression of a dynamic that has existed for years.

2. Structures shape behaviour without determining people

System structures are not invisible machines that completely determine human behaviour.

They alter probabilities.

Rules, budgets, technical interfaces, information flows, hierarchies, deadlines and incentives influence which decisions are easy or difficult, which risks become visible and which behaviours are rewarded.

An individual employee can resist a poor incentive system. But if hundreds of employees operate under the same rules and one metric has influenced compensation or promotion for years, it would be surprising if aggregate behaviour showed no response to that metric.

This helps explain why organisations can continue producing remarkably similar outcomes despite changes in personnel.

New people enter the same processes, receive broadly the same information and make decisions under similar constraints. Personality changes the outcome, but it operates within an existing field of possibilities.

The statement that people respond to structures therefore does not eliminate individual responsibility. It simply means that behaviour cannot be explained through personality alone.

An organisation that repeatedly produces the same conflicts, delays or poor decisions may have more than a personnel problem. It may contain a structure that repeatedly places different people in similar situations.

3. Information structures matter as much as formal rules

A system is shaped not only by what people are allowed to do. It is also shaped by what they are able to know.

Information is rarely distributed evenly across an organisation.

Customer-service employees may see complaints that do not reach product development in a meaningful form until weeks later. A controller may detect rising costs before the operational team understands their cause. Senior management may see highly aggregated indicators while losing precisely those details that first reveal an emerging problem.

These differences are not incidental. They change decisions.

If relevant information arrives late, even a highly capable decision-maker can only react late. If information is heavily compressed, unusual observations may disappear inside an average before anyone recognises that they form a new pattern.

A system can therefore produce poor decisions even when nobody involved is obviously irrational. Each actor decides on the basis of the information available at the time.

Systems thinking consequently asks not only: Who decided?

It also asks: Who knew what — and when?

Information flows can stabilise a system, but they can also make it blind. In many organisations, the problem is not that information does not exist. It exists in the wrong place, arrives too late or is translated into a form that strips away its significance.

4. Feedback makes cause and effect circular

Many intuitive explanations follow a simple sequence: A causes B, and B then leads to C.

Complex systems frequently contain feedback. An outcome alters conditions that later influence the original behaviour again.

The system-dynamics tradition is built around precisely these relationships. Feedback loops, delays and nonlinear responses matter because decisions change their environment, and that changed environment subsequently affects later decisions.^[3]

Two broad forms of feedback are particularly useful to distinguish.

Reinforcing feedback pushes a development further in the same direction. Success may attract more capital, additional capital enables further growth, and that growth makes the company still more attractive to investors.

Balancing feedback pushes against deviation. When inventories fall, a replenishment system may order more goods. As inventories recover, the system reduces new orders.

Both forms can operate at the same time.

A company under financial pressure may cut expenditure. In the short term, its cost position improves.

If training, maintenance or customer service are reduced aggressively, however, productivity, quality or retention may later deteriorate. Results weaken further, creating additional pressure to cut.

What initially appeared to be a response to the problem can become part of a loop that reinforces it.

In such systems, causality cannot be located at one single moment. Each decision changes the conditions surrounding the next one.

5. Delays distort the perception of success

Weeks, months or years may separate a decision from its full consequences.

During that interval, further decisions are already being made. They necessarily rely on an incomplete picture of the original intervention.

A company might lower hiring standards to fill vacancies more quickly. Time-to-hire improves immediately.

Whether onboarding costs, error rates or employee turnover subsequently increase may not become visible for months.

The short-term indicator suggests success while part of the cost remains hidden in the future.

The reverse is equally possible. Investment in training, preventive maintenance or technical modernisation creates an immediate cost even when the benefit emerges only much later.

Systems with long delays therefore create a fundamental evaluation problem: cause and effect may be separated so widely in time that everyday decision-making no longer connects them.

This helps explain why certain measures can remain attractive despite producing weak long-term results. Their benefits appear immediately while their costs arrive later.

Measures with long-term benefits can suffer from the opposite problem. Their costs are visible now, while their value is deferred.

Anyone evaluating only the immediate result may therefore be observing not the system itself, but a temporally distorted fragment of it.

6. The whole can develop properties that nobody planned

A central concept in complexity science is emergence.

Emergent properties or patterns arise through interactions among many components without being fully contained in any single component.^[4][5]

Traffic congestion offers an intuitive example.

No individual driver needs to intend to create a traffic jam. Many independent decisions about speed, following distance, lane changes and braking can nevertheless combine into a persistent traffic pattern.

Similar processes occur in markets, organisations and social networks.

Many locally understandable decisions can produce an aggregate result that none of the participants intended.

A market can become highly concentrated without anyone ever deciding to organise it centrally. A company culture can emerge even though no manual describes it in full. A city can develop a particular traffic problem although no planning decision explicitly aimed at producing that outcome.

Such patterns arise through interaction.

The reverse can also occur: highly complex structures may remain stable even though no individual actor understands the entire system.

Motives therefore explain actions. They do not automatically explain the pattern produced by thousands of connected actions.

This distinction between individual choice and collective outcome is one of the most important insights of systems thinking.

7. Locally rational decisions can damage the wider system

Systemic problems often do not result from obviously irrational behaviour.

They emerge because individual actors optimise their own part of the system rationally.

A purchasing department lowers unit prices by increasing order volumes. Its own metric improves while inventory costs and capital requirements rise elsewhere.

A call centre reduces average handling time. Measured efficiency increases, but unresolved issues may generate more repeat calls.

A sales organisation rewards new customer acquisition more strongly than retention. New business rises in the short term, while churn and service costs may increase later.

None of these decisions needs to be wrong in isolation.

The problem is often the boundary within which success is measured.

Optimising one part of a system can shift costs or risks into another part that lies outside the decision-maker's area of responsibility.

That raises a more fundamental question: Where does the system we are analysing actually end?

The answer is not merely theoretical.

A company can improve one department's efficiency while transferring costs to another. A government can improve one indicator while shifting the burden into another policy area. A production process can become cheaper when environmental or downstream costs fall outside the accounting boundary.

The narrower the boundary, the easier it becomes for an outcome to appear successful.

Systems thinking therefore requires analytical boundaries to be made explicit.

8. What lies outside measurement can still determine the result

Systems are often managed through metrics.

That is unavoidable. Large organisations would be almost impossible to coordinate without measurement.

Every metric, however, compresses reality.

A company may monitor revenue, margin, customer satisfaction, error rates or throughput. Each measure captures something important, but none captures the whole system.

What is not measured does not disappear.

A company may optimise process speed while ignoring the downstream errors created by that acceleration. If it counts only new customers while paying little attention to retention or customer quality, the apparent result will differ from a longer-term assessment.

Measurement can also alter the system it is intended to observe.

Once a metric becomes important for compensation, budgets or promotion, people respond to it.

The metric therefore stops being a passive description of the system and becomes one of the forces shaping behaviour inside it.

That interaction between measurement and behaviour is another reason visible results do not always explain what is actually happening.

9. Stocks explain why systems react slowly

Not everything inside a system changes at the same speed.

Capital, infrastructure, knowledge, reputation, debt, technical debt and trust accumulate over time. Systems theory treats such accumulated quantities as stocks, changed by inflows and outflows.^[1][2]

The classic illustration is a bathtub.

The water level is the stock. The tap and the drain alter it.

Reducing the inflow does not necessarily make the water level fall. As long as more water enters than leaves, the stock continues to rise.

This simple principle explains developments that appear contradictory at event level.

A company may sharply reduce the number of new support tickets while continuing to struggle for months with an old backlog.

A government may reduce its annual deficit while total debt continues to rise.

An organisation may improve recruitment while remaining constrained by legacy systems and accumulated technical debt for years.

Trust follows a similar logic. It may take years to build and a single event to damage, while rebuilding it can take considerably longer.

Changing today's flow does not erase a stock accumulated over many years.

That is one reason systems often respond more slowly than individual interventions suggest they should.

10. Shocks reveal how a system was built before the shock

Crises are commonly narrated through their trigger.

A pandemic, an energy-price shock, the failure of a supplier or a sudden fall in demand appears to explain what follows.

The trigger is real. The consequences, however, depend heavily on the structure already in place.

Two companies can face the same decline in demand and experience radically different outcomes.

One has liquidity reserves, diversified customers and alternative suppliers. The other is heavily indebted, depends on a handful of clients and has been optimised for maximum utilisation.

The external shock may be comparable. The result is not.

Resilience is therefore not simply a property of the event. It arises partly from the structure the event encounters.^[4]

A shock can transform a system.

Just as often, it reveals dependencies, reserves or weaknesses that already existed.

The same principle applies well beyond companies.

Electricity grids, supply chains, financial systems, healthcare services and public administrations can appear stable under normal conditions. Exceptional stress reveals how much redundancy, spare capacity and adaptability they actually contain.

The analytically interesting question is therefore not merely what caused a crisis. It is also why the same pressure was absorbed in one place and amplified in another.

11. Good systems depend less on individual heroes

Organisations often explain success through exceptional people.

There is nothing inherently wrong with this. Experience, talent, leadership and judgement can make enormous differences.

Extreme dependence on a single individual, however, is itself structural information.

If a process works only because one particular person constantly intervenes, remembers undocumented knowledge and improvises around recurring problems, the organisation may not possess a robust process.

It may possess an exceptionally effective human compensation mechanism.

Missing documentation, poor interfaces or unclear ownership remain hidden while that person continues to compensate for them through experience.

The vulnerability only becomes visible when the employee takes leave, changes role or leaves the organisation.

The visible event is then the personnel loss.

The dependency existed long before.

The same pattern can occur at leadership level. An unusually capable manager may compensate for structural weaknesses for years. When problems suddenly become visible after that person's departure, it is tempting to conclude that the successor caused them.

The transition may instead have exposed how much stability previously depended on one individual.

A robust system does not make exceptional people unnecessary. It allows their ability to be used for judgement, improvement and innovation rather than continuous compensation for structural defects.

12. Systems carry their past within them

Many systems were never designed in full.

They grew.

A rule once solved a particular problem. A software platform was built for an earlier generation of products. An organisational responsibility arose because a particular person happened to be there. Infrastructure followed historical transport routes or property boundaries.

Those decisions change the starting point for later decisions.

Systems therefore have histories.

Complexity science studies such path-dependent development precisely because similar starting conditions do not necessarily produce identical outcomes. Historical structures can make later possibilities easier, more expensive or effectively inaccessible.^[7]

The same logic appears inside companies.

Replacing an old software system rarely means replacing software alone. Processes, data structures, skills and business routines may have adapted to it over many years.

A supplier may remain strategically important despite better alternatives because contracts, technical interfaces and internal processes have become aligned with that supplier.

A city may still be shaped decades later by transport routes created under entirely different economic conditions.

What looks inefficient today may therefore be the residue of a solution that once made sense.

Understanding a system often requires reconstructing how it came to exist.

13. System boundaries determine which causes become visible

Every analysis draws a boundary.

That boundary determines what is treated as part of the problem and what is placed outside it.

If only one department is examined, a cost problem may appear to be an internal efficiency issue. Include suppliers, customers and downstream effects, and a different picture may emerge.

Analyse a market only nationally and certain dependencies look different from those visible across an international supply chain.

Evaluate a project over three months and it may appear profitable. Extend the horizon to three years and maintenance costs, customer losses or technical debt may change the conclusion.

There is therefore rarely one objectively perfect system boundary.

What matters is choosing it consciously and recognising which relationships disappear as a result.

This explains many apparent contradictions.

Two analyses can describe the same facts correctly and still produce different conclusions because they include different time horizons, actors or consequences.

Systems thinking is therefore not simply an argument for collecting more data.

It asks the more fundamental question: Which system are we actually analysing?

14. Systemic analysis changes the diagnosis

Systems thinking does not have to begin with a sophisticated mathematical model.

Often, the first step is simply to change the diagnosis.

Instead of treating an incident immediately as an isolated cause, we can ask whether similar outcomes have occurred before.

If a pattern exists, the conditions producing it become relevant. Which information reaches the participants? Which rules apply? Which decisions are rewarded? Which consequences emerge only after a delay? Which stocks have accumulated over years? Where do feedback loops operate?

The choice of system boundary matters as well. If relevant costs, actors or time horizons are excluded, an analysis may remain internally correct while still being incomplete.

The OECD accordingly describes systems thinking as a sense-making approach capable of revealing connections between problems that conventional institutional or professional silos tend to separate.^[4][6]

It does not generate automatic truth.

What it can do is protect against a common analytical mistake: confusing the immediately visible trigger with the complete explanation.

That distinction matters most when the same problem keeps returning.

15. Systems thinking makes responsibility more precise

References to structures carry an obvious risk.

“The system” can become a convenient abstraction behind which personal responsibility disappears.

Systems thinking does not require that conclusion.

People create rules, interpret them, circumvent them and change them. They make concrete decisions and remain responsible within their scope of authority.

That scope, however, is not distributed equally.

An executive who designs a compensation model influences the structure more directly than an employee working within it. A regulator changes different conditions from an individual market participant. A software architect creates dependencies that differ from those faced by the eventual user.

A precise analysis therefore keeps several levels visible at once.

It asks who made a specific decision.

It examines what information was available to that person.

And it asks which structural conditions encouraged, rewarded, tolerated or failed to correct that decision.

Responsibility does not disappear.

It becomes more accurately located.

Events tell stories. Systems explain why stories repeat.

Events remain important. Systems ultimately become visible through concrete decisions, actions and shocks.

The analytical problem begins only when explanation stops at that surface.

Recurring patterns point towards structures. Structures distribute information, create incentives and impose constraints. Feedback carries the effects of earlier decisions into new decisions. Stocks bring the past into the present, while delays push consequences into the future.

System boundaries determine which effects become visible. Information structures shape who can react in time. And many individual actions can combine into outcomes nobody intended.

A single event can transform a system. Just as often, it reveals what was already there: a dependency, a missing buffer, a problematic rule, a distorted information structure or a process whose weaknesses had remained hidden.

Anyone trying to understand why organisations, markets or societies keep producing certain outcomes must therefore look beneath isolated incidents.

That is where the structures lie that turn many individual decisions into a persistent pattern.