Three Sigma, Inc. 

A Systems Thinking Primer
Copyright 2002 Three Sigma, Inc.


Systems thinking is a discipline for applying systems theory to solve real world problems.

A discipline is a body of knowledge, theory, and technique that must be studied and mastered to be put into practice.  

Systems thinking skills can be enhanced through study, coaching, and experience.

Systems thinking is a worldview that sees individuals and organizations as participants in a larger system rather than individual entities reacting to outside forces.  It provides a framework for:

  • Seeing interrelationships rather than things.

  • Seeing patterns of change rather than events.

  • Seeing the structures that underlie complex situations.


Linear Thinking
Focuses on the immediate cause and effect of events.  Cause and effect are assumed to occur together.

Systems Thinking
Focuses on the interrelationship and dynamics among system components. Cause and effect are separated in time and space.

Detail Complexity
Characterized by many variables and complex arrangements. Cause and effect occur together. It is the basis for linear thinking.

Dynamic Complexity
Created by system structural interrelationships and dynamics. Cause and effect are separated in time and space. It is the basis for systems thinking.

Convergent Problems
A quantified and optimal solution is possible. Linear thinking usually provides acceptable solutions to these problems.

Divergent Problems
No best solution can be determined and many solutions are possible. Long-term solutions to these problems usually require a systems approach.

Circles of Causality
Every event or happening is both a cause and an effect.

Examples of circles of causality:










Observed patterns of behavior or results of actions taken.

Highly focused actions that can change system structure.



Systems are entities with consistent patterns of behavior.  They are characterized by virtuous and vicious cycles and oscillating movement toward a stabilizing state or goal. 

Systems thinking consists of identifying the feedback processes and dynamics determining system behavior.

All systems can be modeled using reinforcing (amplifying) processes, balancing processes and delays. 

Reinforcing Processes - engines of growth or decline.

Small actions amplify themselves creating accelerating growth or decline. 

  Example of a reinforcing loop:

Balancing Processes - stabilizing processes that operate whenever there is goal oriented behavior at work. 

All balancing processes contain a self-correcting or governing function that attempts to attain some goal or target. They are characterized by a gap between actual and desired behavior. 

Example of a balancing loop:


System structure causes its behavior.

System interrelationships cause their own crises.  There are no villains.

Understanding these structural interrelationships is necessary to understanding system behavior. 

The ability to influence fundamental change comes from understanding the structures and relationships controlling events and behavior.

Changing system output requires changing the system structure. 

Human systems include the worldview and beliefs of their decision makers and participants.  

Changing the output or behavior of human systems requires changing beliefs.

Learning is the process of changing beliefs. 

Systems inertia creates resistance to change.

Complex systems have high inertia and are very resistant to change. 

Efforts to alter system behavior without changing its underlying structure may create short-term improvements but produce more long-term problems. 

Example: Government elimination of substandard housing and replacing it with subsidized housing projects.

Changing systems structure requires leverage - creating change through highly focused action.

Leverage comes from new ways of thinking.  New ways of thinking include:

Learning to see structures rather than events. 

Thinking in terms of processes of change rather than “snapshots.” 

In human systems, changing the worldview and beliefs of their decision makers redesigns the system.

This potential leverage often goes unrecognized because decision makers usually focus on their own decisions and ignore their impact on others. 

System performance must be optimized at the system level.

Sub-systems both influence and are influenced by the larger system of which they are a part (circles of causality).  This creates internal structural friction within every system.

Minimizing this structural friction requires performance trade-offs and compromise among subsystems. 

When system performance is optimized some sub-systems will be performing at sub-optimal levels. 

Since the entire universe is a system, systems theory applications must first define the boundaries of the system being measured or managed.

Even in a perfect society, some social systems will be functioning at less than optimal levels of performance.

Social responsibility means subordinating some individual desires for the good of society. 

Only one performance measure can be optimized (maximized or minimized) when managing system performance.

All other performance measures become constraints that define the acceptable range of values they can have.

This means that a business cannot maximize both profits and customer value. 

It must choose one to maximize and set constraints on the other.

Example: Maximize customer value (defining how this will be measured) subject to operating profits that provide a minimum after tax return on equity of 20%. 

Determining the optimization measure is the first step in systems management.


Developing a systems perspective

There are multiple levels of explanation for any complex situation. All may be true but their usefulness is different.



Event (Symptoms)


Patterns Of Behavior (trends)


Systemic Structure (root cause)


Event explanations focus on cause and effect. 

They are the most common level of explanation and explain why reactive management prevails.

Pattern of behavior explanations focus on trends and their implications.

They are an attempt to achieve more effective decisions.

Structural explanations are the most powerful and least common. 

They address the root causes of problems where patterns of behavior originate and can be changed.

Creating systems models.

System models are simplified diagrams that capture the essential dynamic complexity of the system under study. 

System models are constructed using combinations of reinforcing loops, balancing loops and delays. 

Well designed systems models will suggest areas of high leverage and low leverage change.

Examples of system models.

 Limits to growth model 

The limits to growth model contains a reinforcing growth loop, a balancing loop that limits growth, and a delay that masks problems.  The following example reflects a growth limiting process that affects every business organization at one time or another.

This model illustrates why organizations have a natural growth rate dictated by some limiting condition.  Problems occur when they try to grow faster than their limiting condition allows.  The failure of many technology companies in 2000 provides an example of organizational growth exceeding financing capability.

Limits to growth situations can have multiple limiting conditions.  In this example the availability of new workers or excessive material lead times could also be limiting factors.

Leverage in limits to growth situations comes from discovering and addressing the limiting conditions or delays.

In this example increasing cash or credit is the high leverage solution. 

Increasing sales demand before addressing the financing limits will only make the problem worse. 

More complex models may have several balancing loops each with their own limiting conditions and delays.  

Shifting the burden model

This model represents a situation many nonprofit and government organizations encounter.  It contains two balancing loops and sometimes a delay that creates undesirable side effects.  In this model one balancing loop represents a quick fix while the second loop presents the long term solution but contains a delay which makes it less attractive.  Here is an example of a shifting the burden model.

In shifting the burden models, the quick solution appears to make the situation better and removes the pressure to pursue the long-term solution.  The side effect makes it even more difficult to invoke the long-term solution because it shifts responsibility for the problem and makes the long-term solution even more difficult to attain.  The result is an over reliance on the quick solution and diminished capacity for the long-term solution. 

This model explains why many well-intentioned efforts to solve a problem only make it worse by enabling the behavior it is trying to correct. 

The leverage in shifting the burden structures lies in limiting the quick solution and strengthening the efforts to accomplish the long-term solution.  Welfare reform legislation is an example of this.

Systems research has identified about 12 generic system models that can be adapted to most situations. For a description of these refer to The Fifth Discipline by Peter Senge, or The Fifth Discipline Fieldbook by Senge and others.



The simplest way to build a system model is to refer to The Fifth Discipline, Appendix 2 and select the generic model that fits the situation under study and use it as a template for model construction.  The Fifth Discipline Fieldbook also provides excellent information on designing system models.

Use the following process to fill in the template and build the model.

Identify problem symptoms or the events creating these symptoms.

This process uses the event level of explanation. 

Look for patterns of behavior or trends that are evident.

This process uses the pattern of behavior level of explanation. 

Patterns of behavior are identified by recurrences of the problem symptoms or related symptoms. 

Use the “multiple why” process to identify the causes underlying these recurring problem symptoms. This process is an adaptation of a Japanese quality technique.  It consists of continually asking “why” to each explanation and subsequent explanations for each of the problem symptoms until a common cause is identified.  

Patterns are identified when problem symptoms can be traced to common causes.

Look for structural relationships creating these problem symptoms.

This process uses the systemic structure level of explanation. 

Continue the “multiple why” process used in identifying patterns of behavior until a fundamental or root cause is apparent.  Structural relationships are identified when the explanation for the problem symptom changes from one system component to another, i.e., explanation for homelessness moves from society (unemployment) to the individual (addiction) or when the explanation for a quality problem moves from manufacturing (defective product) to procurement (improper material).

The system components identified in this process become the entries on the reinforcing and/or balancing loops in the template selected at the beginning of this process.

Fill in the feedback loops and delays.

Reinforcing loops (virtuous spirals or vicious circles).

These will be indicated when activity reflects growth or decline or when problem symptoms get better or worse.

Balancing loops

These are the most common and reflect a search for stability. These loops are present when activity or problem symptoms involve capacity limitations, goals or targets. These loops will be characterized by gaps between what exists and what is desired.


Most systems involve combinations of reinforcing and balancing loops. 

All reinforcing loops will have one or more balancing loops limiting growth or decline. 

All balancing loops will have a limiting or governing function which regulates its output within the parameters defined by some desired or limiting function. High leverage actions will involve changing the parameters defining the limits regulating this balancing process. 

Locate the delays separating cause and effect.

Understanding the delays in seeing the results of actions taken will prevent over-controlling or excessive actions that create system instability rather than change. 

Use the completed model to discover the points of leverage.  

The template selected at the beginning of this process will help identify the potential points of leverage.

Return to Analytical Tools and Models