The first management decision in a difficult situation is not what to do. It is deciding what kind of system you are dealing with.
A technically difficult initiative can still be predictable. A genuinely complex initiative may look manageable on a plan while behaving in ways that no schedule can fully anticipate. Confusing the two is one of the most expensive mistakes leaders can make because the error sits upstream of almost every later decision.
A difficult engineering task may require exceptional expertise, detailed analysis and disciplined execution, yet still have stable requirements and discoverable cause-and-effect relationships. A complex transformation may involve familiar technology but uncertain stakeholder responses, interacting dependencies, changing priorities and feedback loops that alter the system while it is being managed.
The distinction matters at executive, portfolio and program level. If leaders treat complexity as nothing more than greater difficulty, the normal response is more planning, more controls, more reporting and more escalation. Those interventions can help when the work is complicated. In a complex system, they can also slow learning, suppress local judgement and create an illusion of control while the underlying system keeps changing.
The Strategic Context
The supplied research does not offer a single universal definition of project complexity. San Cristóbal's 2017 review explicitly notes the lack of consensus and shows that different models emphasise different dimensions: goals and methods, agreement and certainty, the number of elements, interdependencies, technology, organisational structure and uncertainty.
Oehmen, Thuesen, Parraguez and Geraldi take a systems-oriented position. Their 2015 PMI white paper distinguishes structural complexity, which concerns the composition and relationships of the system, from dynamic complexity, which concerns how the system behaves and changes over time. It also separates simple, complex and chaotic conditions because different conditions require different managerial responses.
That distinction is strategically useful. It shifts the leadership question from "How hard is this?" to "How predictable is cause and effect, how stable are the relationships, and how quickly can we learn?"
This matters because portfolios and programs routinely contain a mixture of conditions. A regulatory compliance project may be highly complicated but relatively predictable. A new digital service may have technically familiar components but uncertain customer adoption. An enterprise operating-model transformation may combine both with political ambiguity, resource competition and behaviour that emerges only after implementation begins.
A portfolio that treats all three in the same way is not standardised. It is misdiagnosed.
What Leaders Commonly Misread
The first misread is to equate size with complexity. Large budgets, long schedules and large teams can increase exposure, but a large project is not automatically complex in the systems sense. Conversely, a small initiative can become highly complex if it touches many tightly coupled processes or influential stakeholders.
The second misread is to assume that more detail creates more certainty. Detailed plans are valuable when relationships are stable enough for the detail to remain meaningful. When assumptions, interfaces or stakeholder expectations are changing, a highly detailed baseline can become a precise description of a world that no longer exists.
The third misread is to assume that failure to predict means failure to manage. Complex conditions often make precise prediction weaker, but management can still improve outcomes through shorter learning cycles, explicit assumptions, modular design, adaptive decision rights and stronger sensing.
The fourth misread is to make "agile" the automatic answer. The Week 11 teaching material correctly separates organisational agility from the formal use of Agile delivery methods. The relevant question is not whether teams use sprints. It is whether the decision system can respond appropriately to what is being learned.
Reframing the Issue
Complexity should be treated as a management design variable.
Before leaders choose controls, they should diagnose the decision environment. Oehmen and colleagues' distinction between simple, complex and chaotic systems is useful precisely because it does not assume one management model is universally superior. In simple conditions, proven practice and clear standards may be sufficient. In complex conditions, analysis, experimentation and learning matter more. In chaotic conditions, the first requirement may be stabilisation rather than optimisation.
The Cynefin material in the supplied Study Notes makes a similar point: managers can create additional problems when a preferred management style is imposed on a context for which it is poorly suited.
This reframing has an important consequence. "Best practice" is conditional. A control that is excellent in a stable process can be harmful in an emergent transformation if it prevents adaptation. A decentralised decision that accelerates learning in a digital product team can be unacceptable in a tightly regulated safety decision. A single governance template therefore cannot substitute for diagnosis.
Diagnose the Decision Environment Before Choosing the Method
Leaders can start with four questions.
1. How stable is cause and effect?
If cause and effect are well understood, analytical planning and standard controls have strong value. If relationships are delayed, nonlinear or shaped by feedback, the plan should include learning mechanisms rather than assume that the initial model is complete.
2. How much agreement exists about the outcome?
San Cristóbal's discussion of Stacey's Agreement and Certainty Matrix is useful here. A technically understandable solution does not eliminate complexity if powerful stakeholders disagree about what should be achieved. In that case the constraint may be negotiation and coalition-building rather than engineering knowledge.
3. How tightly coupled are the components?
A portfolio containing many independent initiatives may be easier to govern than a smaller portfolio in which projects share scarce experts, data, platforms, suppliers and executive decisions. Interdependency, not component count alone, drives much of the structural complexity.
4. How quickly can the system change relative to our ability to observe it?
This is the boundary between manageable complexity and practical chaos. If the environment changes faster than leaders can sense, interpret and respond, adding analysis may not help. Stabilising constraints, protecting critical outcomes and using robust decision rules may be more important than producing a more elaborate forecast.
Decision Framework
A useful executive diagnostic is to classify the work across five dimensions before approving the management approach.
| Dimension | Lower-complexity condition | Higher-complexity condition | Leadership implication |
|---|---|---|---|
| Cause and effect | Known and repeatable | Delayed, nonlinear or emergent | Increase learning and feedback |
| Stakeholder agreement | Broadly aligned | Competing interpretations or interests | Invest in sensemaking and negotiation |
| Interdependence | Limited, modular | Dense, cross-functional coupling | Map interfaces and shared constraints |
| Uncertainty | Information can reasonably close gaps | Important unknowns remain | Test assumptions and preserve options |
| Rate of change | Slower than decision cycle | Faster than governance response | Shorten decision cycle or stabilise scope |
The purpose is not to calculate a complexity score and declare victory. It is to expose where the management model may be mismatched.
A complex initiative with low stakeholder conflict but high technological uncertainty may need experimentation and technical modularity. A program with known technology but high political disagreement may need stronger sponsorship, negotiation and decision-right clarity. A chaotic operational incident may need immediate containment and a small set of robust priorities before detailed analysis begins.
Different problems require different combinations of governance, evidence and authority.
The Portfolio Implication: Balance Complexity, Do Not Eliminate It
Portfolio leaders face an additional problem. Avoiding all complex work would also avoid many sources of innovation and strategic renewal.
The supplied Study Notes make this tension explicit when discussing certainty and agreement. Restricting a portfolio to work close to certainty can reduce execution risk, but it can also prevent the organisation from pursuing high-reward opportunities such as new technologies or new business models.
The portfolio objective is therefore not minimum complexity. It is deliberate exposure to complexity that the organisation has the capability and capacity to absorb.
That requires leaders to ask not only whether each initiative is attractive, but whether too many initiatives depend on the same uncertain technology, the same executives, the same suppliers or the same organisational change capacity.
Complexity concentration can become portfolio risk even when each business case looks reasonable in isolation.
Related article: Portfolio Management Is Not a Rational Optimisation Problem
From Strategy to Execution
Immediate action should begin with diagnosis. For major initiatives, add a short complexity assessment before deciding the delivery and governance model. Identify unstable assumptions, critical interfaces, contested outcomes and the rate at which the environment can change.
Medium-term capability should focus on management flexibility. Build governance patterns that can vary decision cadence, escalation thresholds, experimentation authority and evidence requirements according to context. Train leaders to recognise when a problem has moved from complicated to complex, or when an apparently complex problem can be simplified through modularity.
Long-term positioning requires the organisation to become better at learning. This means preserving evidence from decisions, reviewing assumptions rather than only milestones, developing leaders who can operate without false certainty and designing portfolios so that uncertain bets do not all fail through the same dependency.
The objective is not to predict everything. It is to prevent the organisation from being surprised by the fact that prediction has limits.
Signals to Monitor
Watch for these indicators that the management model may no longer fit the system:
- plans are repeatedly rebaselined while the underlying assumptions remain unexamined;
- reporting volume rises but decision quality does not improve;
- local teams wait for escalations on issues they understand better than the governing body;
- new controls create longer queues, workarounds or shadow decision channels;
- apparently independent projects begin competing for the same scarce capability;
- stakeholder agreement deteriorates even though technical progress remains on plan;
- small changes create disproportionate consequences elsewhere in the system.
These are not proof of complexity. They are signals that leaders should re-diagnose rather than simply tighten control.
Questions for the Leadership Team
- Which parts of our portfolio are genuinely complex, and which are simply difficult or technically demanding?
- Where are we applying a management method because it is familiar rather than because it fits the decision environment?
- Which assumptions would make our current plan invalid if they changed?
- Where does stakeholder disagreement matter more than technical uncertainty?
- Which initiatives share dependencies strongly enough to create portfolio-level complexity?
- What decisions could safely be decentralised to shorten our learning cycle?
- If conditions became chaotic, what would we protect first and what could we temporarily stop?
Closing Perspective
Complexity does not excuse weak management. It demands better management judgement.
The leadership task is not to choose between discipline and adaptability. It is to understand when discipline should create stability, when analysis should deepen understanding, when experimentation should accelerate learning and when decisive containment is required.
A plan is only as useful as the model of reality behind it. Leaders who diagnose the system before managing the work are better positioned to decide what deserves control, what requires learning and what must remain flexible.
Related article: From Framework Knowledge to Executive Judgement: Diagnose Before You Recommend
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.