When work becomes hard, the usual response is more of the same management: more planning, more detail, more reporting, more control, another options paper. Sometimes that is exactly right. Sometimes it makes things worse, because the problem is not the kind that more planning can solve.
A technically demanding job, such as a precision machining order or a complex electrical installation, can be difficult but predictable: the requirements are stable and cause and effect can be worked out by experts. A business change, such as introducing a new system or entering a new market, may involve familiar technology but unpredictable reactions from customers and staff, shifting priorities and effects that only appear once work begins. And some situations are unclear not because information is missing but because the people involved want different things. Each kind of problem needs a different approach.
This article explains how to tell difficult, complex, chaotic and ambiguous situations apart, why treating them the same way causes trouble, a short diagnosis to run before choosing how to manage, and how to handle disagreement that no amount of data will resolve.
Difficulty is not the same as complexity
It helps to distinguish a few kinds of situation, drawing on frameworks such as the Cynefin framework developed by Dave Snowden and colleagues, and Ralph Stacey’s agreement and certainty matrix:
| Situation | Cause and effect | What works |
|---|---|---|
| Clear | Known and repeatable | Standard procedures and good practice |
| Complicated (difficult) | Knowable with expertise and analysis | Expert analysis, detailed planning, disciplined execution |
| Complex | Only clear in hindsight; shaped by interactions and feedback | Small experiments, short feedback loops, learning as you go |
| Chaotic | No stable pattern; things changing too fast to analyse | Act quickly to stabilise, then reassess |
A large job is not automatically complex. A small change can be highly complex if it touches many tightly connected processes or influential people. And research on project complexity has not settled on a single definition; different models emphasise the number of elements, how tightly they interact, uncertainty, technology, organisational structure and stakeholder disagreement. The useful point is not the label but the question: how predictable is cause and effect here, how stable are the relationships, and how quickly can we learn?
Why the wrong approach hurts
If a complex situation is managed as if it were merely difficult, the natural response to problems is more detail and control. Detailed plans are valuable when relationships are stable enough for the detail to stay true. When assumptions, connections or people’s reactions keep changing, a detailed plan becomes a precise description of a situation that no longer exists. More controls can slow learning, discourage people from using judgement and create a false sense of control.
The opposite mistake also happens. Treating a difficult but predictable job as an experiment wastes the expertise and planning that would have delivered it reliably.
Two other misreadings are common:
- Failing to predict is not the same as failing to manage. In complex situations, good management means shorter learning cycles, explicit assumptions, modular designs and the ability to adjust, rather than better forecasts.
- “Agile” is not automatically the answer. The question is not whether a team uses a particular method, but whether the business can respond sensibly to what it learns.
Missing information or competing interests?
A separate distinction matters just as much. Some unclear situations are uncertain: information is missing, and getting it would narrow the choice. Others are ambiguous: the information may be adequate, but people interpret it differently or want different outcomes.
More data reduces uncertainty. It does not settle a disagreement about values, priorities or who bears the cost. A business can commission analysis after analysis to avoid a choice that is actually about competing interests, and the decision will still be waiting at the end.
Stacey’s matrix combines the two dimensions:
| Agreement | Certainty | Situation | Useful approach |
|---|---|---|---|
| High | High | Clear goal, known method | Plan and execute |
| Low | High | The method is understood, but people disagree about the outcome | Negotiation and clear decision rights |
| High | Low | Everyone agrees on the goal, but nobody knows which method will work | Experiments, pilots and staged commitments |
| Low | Low | Disagreement and unpredictability together | Make sense of the situation together, keep options open, break the problem into smaller decisions |
Situations move between these as evidence, people and circumstances change.
Two kinds of ambiguity
Some ambiguity is avoidable: it comes from vague purpose, inconsistent language, unclear responsibilities or unstated assumptions. Remove it quickly. Clear purpose and decision rights cost little and help everything else.
Some ambiguity is irreducible: legitimate interests genuinely conflict, or the future is genuinely open. This cannot be documented away. It has to be governed: by making the trade-offs explicit, deciding who decides and agreeing what would cause the decision to be revisited.
Two failures sit at either end. Forcing a single interpretation too early can shut down useful options before the problem is understood. Leaving ambiguity unresolved too long leads to endless consultation. The skill is knowing what must be decided now and what can stay open while the business learns.
Making sense together
When people interpret a situation differently, the leader’s job is to help build enough shared understanding to act. This is sometimes called sensemaking. In practice it involves:
- making assumptions explicit;
- separating facts from interpretations;
- identifying where people genuinely disagree, and where different words hide similar concerns;
- showing who bears the costs, risks and benefits of each option;
- distinguishing reversible decisions from hard-to-reverse ones;
- agreeing what new evidence would change the choice.
The aim is not unanimity. It is enough shared understanding of purpose, trade-offs and decision authority for people to act together.
A short diagnosis before choosing how to manage
Before deciding how to run a significant piece of work, ask six questions:
- How predictable is cause and effect? If it is well understood, plan in detail. If effects are delayed, indirect or shaped by people’s responses, build in learning.
- How much agreement is there about the outcome? If key people disagree, the main task may be negotiation and decision rights, not analysis.
- How tightly connected are the parts? Work that shares people, equipment, data or suppliers with other work is more complex than its size suggests.
- Would more information genuinely narrow the choice? If yes, get it. If not, stop treating the problem as analytical.
- How fast is the situation changing compared with how fast we can respond? If faster, simplify, stabilise and protect what matters most.
- Who has authority to decide? Ambiguity becomes paralysis when nobody knows.
The result is not a score. It is a judgement about which parts of the work need tight planning, which need experiments, which need negotiation and which need a quick stabilising decision. Many initiatives contain all four.
When things become chaotic
Occasionally a situation changes faster than anyone can analyse it: a major equipment failure during peak season, a key customer collapsing, a flood, a cyber incident. In these moments, careful analysis is the wrong first step. The priority is to stabilise: protect people, protect the most important customers and commitments, stop the damage spreading and establish a few clear priorities. Decisions are made quickly by whoever is in charge, and communicated simply.
Once the situation is stable, the business can move back to analysis and learning, and review what happened. The mistake is staying in emergency mode after the emergency has passed, with decisions still centralised and routines still suspended. Agree in advance who leads in a crisis and what would mark its end.
Do not avoid all complexity
Avoiding all complex work would also avoid most innovation and renewal. New products, new markets and new ways of working are complex by nature. The aim is not to eliminate complexity but to take on the amount the business can absorb, and to avoid having several uncertain initiatives all depending on the same few people, the same supplier or the same untested technology.
A worked example
This is an illustration. A wholesale bakery supplying cafés is running three initiatives at once, all managed the same way: a detailed project plan, a fixed budget and fortnightly status reports.
- Replacing an oven: a difficult, technical job with clear requirements, an experienced supplier and a fixed installation window.
- Launching an online ordering portal for cafés: familiar technology, but nobody knows how cafés will actually use it or what they will ask for once they do.
- Reorganising delivery runs: drivers, the sales manager and the owner disagree about whether to prioritise early delivery times or fewer kilometres.
After four months, the oven is installed on schedule. The portal plan has been rebaselined three times, and the delivery reorganisation has produced two analyses and no decision.
The owner runs the diagnosis:
- The oven is complicated but predictable. Detailed planning was right, and it worked.
- The portal is complex: café behaviour can only be learned by watching it. The owner replaces the detailed plan with a short cycle: release a basic version to ten cafés, review usage and feedback fortnightly, and add features only when there is evidence they are needed.
- The delivery runs are ambiguous, not uncertain. More analysis will not resolve a disagreement about what matters most. The owner brings the three parties together, sets out who bears the cost of each option, agrees that early delivery for the twenty largest cafés is non-negotiable and the rest can be optimised for distance, and names the operations manager as the decision-maker for future changes, with a review in three months.
The portal settles into a steady improvement rhythm, and the delivery decision, which analysis could not make, is made in a single meeting.
How this applies to a small Australian business
Small businesses often run very different kinds of work through the same habits. Practical steps:
- Diagnose before choosing how to manage: clear, complicated, complex, chaotic or ambiguous.
- Plan in detail where cause and effect are well understood.
- Use small experiments and short feedback loops where they are not.
- Separate missing information from competing interests.
- Remove avoidable ambiguity quickly with clear purpose and decision rights.
- Govern irreducible ambiguity with explicit trade-offs and a named decision-maker.
- Avoid stacking several uncertain initiatives on the same people or suppliers.
The articles on plans that detect rather than predict and what your methods cannot see cover related approaches.
Signals worth watching
- Plans rebaselined repeatedly while assumptions go unexamined.
- More reporting with no better decisions.
- Decisions deferred for “more information” that never changes anyone’s view.
- People using the same words but meaning different outcomes.
- Agreement in meetings followed by different priorities afterwards.
- Small changes causing large, unexpected effects elsewhere.
- Several uncertain initiatives relying on the same few people.
Common mistakes
- Equating size with complexity.
- Answering complexity with more detail and control.
- Treating predictable work as an experiment.
- Commissioning analysis to avoid a decision about competing interests.
- Forcing one interpretation too early, or never deciding at all.
- Leaving decision rights unclear in ambiguous situations.
Frequently asked questions
How do I know whether a problem is complex or just difficult? Ask whether an expert could work out the answer in advance with enough analysis. If yes, it is difficult. If the answer only becomes clear by trying things and seeing how people and systems respond, it is complex.
Does complex work need no plan at all? It needs a different kind of plan: a clear goal, explicit assumptions, small steps, frequent reviews and agreed points to change course.
What if the disagreement is between the owners? The same principles apply: separate facts from preferences, make trade-offs explicit and agree how the decision will be made. An outside facilitator or adviser can help.
How often should we re-diagnose? Whenever results surprise you, people’s positions shift or the situation changes quickly. Work can move from complex to complicated as you learn, or the other way.
Is this just theory for large organisations? No. Small businesses face the same mix of predictable jobs, uncertain innovations and disagreements, often with fewer people to absorb the cost of handling them the wrong way.
Can the same initiative be several types at once? Usually. A new product launch might combine a complicated manufacturing setup, a complex question about customer demand and an ambiguous disagreement about pricing. Manage each part in the way that suits it, rather than forcing the whole initiative into one approach.
Questions to ask
- Which of our current initiatives are predictable, and which are genuinely uncertain?
- Where are we using detailed plans for work that can only be learned by doing?
- Which disagreements are about missing information, and which are about competing interests?
- What ambiguity could we remove today with clearer purpose or decision rights?
- Who has the authority to decide where we disagree?
- Are several of our uncertain bets depending on the same people or suppliers?
Bringing it together
Not all hard problems are alike. Predictable but difficult work rewards careful planning and expertise. Complex work rewards small experiments and fast learning. Chaotic situations need stabilising action first. Ambiguous situations need negotiation, clear decision rights and shared understanding, because more data will not settle a disagreement about what matters. Diagnose the kind of problem before choosing how to manage it, and expect a single initiative to contain more than one kind.
Source: KEVOS notes, drawing on teaching material and published reviews of project complexity, including the Cynefin framework, Stacey’s agreement and certainty matrix and research distinguishing uncertainty from ambiguity. Examples and figures in this article are illustrations.
