System Boundaries: How to Decide What Belongs in Your Systems Analysis

A system boundary separates what an analysis treats as inside the system from what it treats as external. Boundaries are necessary: no map or model can include everything. But they are not neutral. A boundary determines which causes, stakeholders, costs, time horizons, and interventions become visible.

Good systems work therefore treats a boundary as an explicit analytical choice, not as an objective line waiting to be discovered.

Why boundaries matter

Consider a delivery company optimizing route time. A narrow boundary may include drivers, vehicles, and stops. A wider one may include customer time windows, warehouse decisions, subcontractor incentives, traffic, emissions, and worker fatigue. Each boundary produces a different account of performance.

A narrow model is not automatically wrong. It may answer a specific operational question efficiently. The problem occurs when conclusions are generalized beyond the boundary or when excluded effects are treated as if they do not exist.

Five dimensions of a boundary

• Purpose: which decision must the analysis support?
• Actors: whose behavior and perspective are included?
• Mechanisms: which feedback loops, rules, and resources matter?
• Time: how far backward and forward does the analysis look?
• Place and scale: which organizational, geographic, or institutional levels are included?

A practical boundary-setting process

1. Write the focal question and intended decision.
2. List directly affected groups, including those with limited decision power.
3. Sketch the immediate process and its inputs and outputs.
4. Follow important effects across departments, places, and time.
5. Identify external factors that behave as constraints versus variables that the intervention can influence.
6. Create a primary working boundary and a list of monitored exclusions.
7. Ask what new evidence would require the boundary to expand or contract.

Boundary critique and power

Boundary choices decide who counts as a stakeholder, what counts as expertise, and which harms count as costs. Critical systems traditions emphasize asking who benefits from the framing, who bears consequences, and who lacks influence over the analysis.

This does not require an infinitely large model. It requires transparent justification and a willingness to compare alternative boundaries.

Test sensitivity to the boundary

Run the analysis under at least two plausible boundaries. Does a recommended intervention still look attractive when delayed effects are included? Does efficiency disappear when rework is counted? Does a local improvement create costs for another team or population?

If the recommendation changes substantially, the boundary is a major source of uncertainty and should be reported as such.

Common boundary mistakes

• Equating the organizational chart with the real system.
• Including only groups with formal authority.
• Choosing a time horizon shorter than important delays.
• Treating environmental or social costs as external and therefore irrelevant.
• Expanding the map until it can no longer support a decision.

References and further reading

• Ulrich, W. (1983). Critical Heuristics of Social Planning. Haupt.
• Midgley, G. (2000). Systemic Intervention. Springer.
• Churchman, C. W. (1970). “Operations Research as a Profession.” Management Science, 17(2), B37–B53.
• UK Government Office for Science, Systems Thinking Toolkit.