Systems Theory

Systems Thinking Toolkit

1960s. Jay Forrester at MIT builds the first computer model of Boston's urban dynamics. The simulation predicts: subsidized housing will destroy entire neighbourhoods within 20 years. The policymakers don't believe it - and get exactly that result. Forrester didn't guess the future. He simply drew the right diagrams and followed the feedback loops to their conclusion.

Business process analysis
Diagnosing organizational problems
Strategic planning
Retrospectives and post-mortems

Causal Loop Diagrams: Seeing Connections

**The problem:** Explaining a complex situation to colleagues. 'A affects B, B affects C, but C also affects A, and there's a delay...' A minute in, everyone has lost the thread.

**The solution:** Causal Loop Diagram (CLD) - a visual language for cause-and-effect relationships.

**CLD** - a diagram showing variables (what changes) and connections between them (how one thing affects another). It allows feedback loops to become visible.

**Example: Viral product growth**

**Common mistake:** Drawing arrows between events instead of variables. 'Launch campaign → sales growth' is an event. Correct: 'Ad budget (+)→ Brand awareness (+)→ Sales'

In a CLD, an arrow with (+) means...

Stock-Flow Mapping: Bathtubs and Faucets

A CLD shows connections, but doesn't show **accumulations**. For that a Stock-Flow Diagram is needed.

**Stock** - an accumulation, a reserve (a bathtub of water). **Flow** - a stream that changes the stock (faucets). Stock is measured in units; Flow in units per time.

**Key insight:** Stock changes SLOWLY. Even if all the faucets are turned off - the water in the bathtub doesn't disappear instantly.

Stock	Inflow	Outflow	What people miss
CO2 in atmosphere	Emissions	Absorption	CO2 accumulates over decades
Reputation	Positive reviews	Scandals	Reputation takes years to build
Technical debt	Quick hacks	Refactoring	Debt accumulates invisibly
Team trust	Keeping promises	Violations	Takes longer to rebuild than to lose

**Why this matters:** Managers often affect flows (hiring/firing), but expect immediate changes in stocks (team productivity). Stocks have inertia!

A company rapidly increased hiring. Why won't productivity improve immediately?

The Iceberg Model: Digging Deeper

What is visible are **events** - the tip of the iceberg. But the causes lie deeper: in patterns, structures, and mental models.

**Iceberg Model** - a tool for moving from reactive thinking (responding to events) to systems thinking (changing structures).

Applying this to an employee who is always late

**Event:** Ivan was late to a meeting. **Pattern:** Ivan is late 2-3 times a week. **Structure:** Meetings are scheduled back-to-back with no transition time. No consequences for being late. **Mental model:** 'A meeting starts when everyone has arrived' (company culture).

Level	Question	Type of action
Events	What happened?	Reactive (fighting fires)
Patterns	What keeps repeating?	Adaptive (preparing)
Structures	Which connections create the pattern?	Proactive (changing the system)
Mental models	Which beliefs drive this?	Transformational (changing culture)

**The trap:** Most organizations get stuck at the events level. 'Ivan was late - let's talk to Ivan.' This doesn't solve the systemic problem.

A project is missing its deadline again. Which question helps move to the structures level?

Behavior Over Time: Dynamic Graphs

**BOT (Behavior Over Time)** - the simplest yet most powerful tool. Draw how a variable changes over time.

**BOT Graph** shows the dynamics of a variable. The X-axis is time, the Y-axis is value. The graph's pattern reveals the type of system behavior.

**How to use BOT:**

Choose 3-5 key variables in the system
Draw their behavior over the past period
Ask: 'What's the pattern?' - growth, decline, oscillations, S-curve?
Ask: 'What structure creates this pattern?'

BOT for a startup

Variables: Users, Revenue, Burn Rate, Runway. If Users grows exponentially but Revenue grows linearly - that's a problem. If Burn Rate grows faster than Revenue - runway is shrinking. Graphs make this immediately visible.

A variable grows at first, then slows down and levels off at a plateau. What pattern is this?

Full System Map: Putting It All Together

Now all the tools combine into a **System Map** - a complete picture of system understanding.

**System Map** combines: BOT (what's happening?), Iceberg (why?), CLD (what connections?), Stock-Flow (where are the accumulations?).

**Example: Team attrition**

**BOT:** Attrition has been increasing over the last 6 months
**Iceberg:** Event - Maria left. Pattern - 3rd departure this quarter. Structure - no growth path, overload. Mental model - 'people are replaceable'
**CLD:** Overload (+)→ Burnout (+)→ Departures (+)→ Overload on remaining staff (R-loop!)
**Stock-Flow:** Stock = experienced employees. Outflow > Inflow. Stock is depleting.
**Leverage:** Break the R-loop - reduce workload BEFORE more people leave.

**Common mistake:** Starting with CLD (complex). Better to start with BOT (simple) and Iceberg (structures thinking).

Systems analysis requires complex software and mathematics

Paper, a pen, and the right questions are enough

The tools are ways of organizing thinking. Draw a BOT on a napkin, ask 'what's the pattern?' and 'what structure lies behind it?' - that is already systems analysis.

Which tool is best to START a systems analysis with?

The Systems Thinking Toolkit

CLD - visualizing cause-and-effect connections and feedback loops
Stock-Flow - accumulations and flows, system inertia
Iceberg - from events to patterns, structures, and mental models
BOT - graphs of variable behavior over time
Start simple (BOT), add complexity as understanding grows