Managing complexity: what VUCA leaders get wrong

Managing complexity has produced the most quoted piece of bad advice in business. Faced with a complex environment, simplify. It sounds sensible, it sells well, and it is the reliable way to make a complex system more dangerous rather than less.

Saying the world has become complex is now a cliche. Constant connection, endless information and instant global reach do give the impression that complexity is growing faster than anyone can absorb it.

The world has always been complex. Human organisations have been complex since people started forming groups, and no technology will ever reduce the complexity your business faces. Understanding it well enough to use it is a different proposition, and that one is available.

Here is the observation I want to add, drawn from doing this work rather than from reading about it. Ask an executive what makes their business complex and you get regulation, competition, geography and technology. Ask the people doing the work and you get obsolete processes, unclear responsibilities and contradictory instructions. Those two answers describe different problems, and only one of them is inside your control.

This article continues my series. It follows VUCA and the management of uncertainty, and the reference page for the cluster is my pillar on VUCA leadership and where decisions are made.

The definition	The reflex it triggers	The measure that changes the diagnosis
Managing complexity, definedManaging complexity means leading an organisation whose parts interact in ways nobody can fully trace. VUCA complexity is the difficulty of understanding those interactions, and of predicting the primary and downstream effects of changing any one of them. Interconnection replaces clean cause and effect, so a small change can produce large consequences whose origin cannot be traced.	Simplify, which rarely worksThe instinctive response to a complex system is to simplify it. Removing genuine complexity is valuable where it creates no value, and hiding it merely relocates the surprise to a later date. A complex system cannot be understood by any individual, which makes the useful response collective rather than reductive.	Measure the complexity your people carryExecutives describe complexity that arrives from outside, through regulation, competition and technology. The people doing the work describe complexity that arrives from inside, through obsolete processes, unclear responsibilities and contradictory instructions. The second kind is designed rather than inherited, and it is the kind an organisation can act on this quarter.

What VUCA complexity actually means

VUCA complexity is the difficulty of understanding the interactions of many parts or factors, and of predicting the primary and downstream effects of modifying any of them within a highly interdependent system. Branches and non-linear consequences multiply so fast that they overwhelm most assessment processes, which is how complexity generates uncertainty.

That definition is, appropriately enough, the most complex of the four. Let me pull one word out of it, because the word carries the whole idea.

Interconnection is the operative word

Interconnection is what turns a large system into a complex one. When everything touches everything else, a minor change can produce major consequences whose cause cannot be traced backwards. Causality gives way to a tangle of reactions and counter-reactions, and choosing the right path becomes close to impossible.

A system is called complex when its behaviour cannot be described from knowledge of its components and their interactions alone. That is a precise technical claim rather than a figure of speech.

It means the whole does things the parts do not explain, so no amount of studying the parts will get you there. This is why complexity resists the analytical habits that work everywhere else in management.

It also means a complex system never repeats its behaviour exactly. You cannot rerun a historic event to get a better photograph of it, which removes the possibility of learning by controlled repetition.

Complicated and complex are different problems

A complicated system has many parts and a knowable answer. A complex system has many interacting parts and no reliably knowable answer. A mechanical watch is complicated, since hundreds of components produce a perfectly anticipated movement that an engineer can explain at any moment.

Run that watch a thousand times and you get the same result a thousand times. Every interaction is known, intended and controlled, and the same holds for planetary motion.

Now take drug use in a population. There are sociological explanations, economic explanations and geopolitical explanations, and the relevant variables can be multiplied indefinitely. Some people read it as a criminal problem and others as a public health problem, and both readings are defensible.

Decades of enormous resources spent on that problem with little measurable movement is what a complex system looks like when it is treated as a complicated one.

How complexity differs from uncertainty

Uncertainty is about not knowing what will happen. Complexity is about not being able to work out why something happened even after it has. You can forecast the immediate result of a single interaction inside a larger network, and the branches multiply so quickly that the overall behaviour escapes you.

The relationship runs one way. Complexity generates uncertainty through the sheer volume of possible interactions and outcomes, and reducing uncertainty does nothing to reduce complexity.

That ordering explains the sequence of the letters. Volatility feeds uncertainty, uncertainty compounds into complexity, and the acronym reads in the direction the problem travels.

It also explains why the responses differ. Uncertainty rewards cheap exits, which I cover in the uncertainty article, and complexity rewards collective seeing, which is where this one is heading.

The Cynefin framework and what each context asks of a manager

Cynefin is a sense-making framework developed by David Snowden that sorts situations into distinct decision contexts, each calling for a different response. Published with Mary Boone in Harvard Business Review, it separates clear and complicated contexts, where cause and effect can be established, from complex and chaotic ones, where they cannot.

I use it as the shallow end of the pool for teams meeting complexity for the first time, because it converts an abstract idea into a question a manager can answer in a meeting.

Clear and complicated, where analysis still works

Clear contexts are stable, with cause and effect visible to everyone and usually one right answer that already exists in a process. Complicated contexts have a genuine causal relationship that requires expertise to see, and several defensible answers. Both reward assessment, categorisation and the application of known practice.

Snowden and Boone describe the response in clear contexts as sensing the situation, categorising it and responding with established practice, which is exactly how a well-run service desk operates.

The danger in clear contexts is oversimplification. Leaders who have succeeded repeatedly start applying the same solution automatically, and stop noticing situations that fit no existing category.

The danger in complicated contexts is different. Over-reliance on experts crowds out creative answers from people outside the specialism, which is why I insist on mixed groups that include the sceptics and the awkward questioners.

Complex and chaotic, where analysis stops working

In a complex context there may be no correct answer and no traceable cause and effect, so the sequence becomes probe, sense and respond. In a chaotic context there is no relationship between cause and effect at all, and the first job is to establish enough order to move the situation into complexity.

The complex response is patient and collective. Look for patterns, run experiments, allow a solution to emerge from a group, and treat failure as part of the process rather than as a fault to attribute.

Snowden and Boone argue that the complex domain is far more common in business than most leaders realise, and that it requires responses that often feel counterintuitive.

The chaotic response inverts that entirely. Act first, sense where stability is appearing, then respond, because in genuine chaos deliberation is a luxury the situation does not offer.

Disorder, and the cost of misreading the context

The framework includes a fifth state, disorder or confusion, where it is unclear which of the other contexts applies. The reflex in that state is to fall back on familiar decision tools such as decision trees and priority matrices, which quietly assume an ordered world. The first job is to gather enough information to identify the context.

Misreading the context is the expensive error, and it runs in both directions. Treating a complex problem as complicated produces an elaborate plan that fails for reasons nobody predicted.

Treating a complicated problem as complex produces endless experimentation on a question an expert could have settled in a morning.

The framework comes with a warning worth repeating in full to any leadership team. Believing that everything is simple and orderly, and that past success makes you safe from future failure, is itself the largest problem in the room.

Where organisational complexity actually comes from

Complexity inside an organisation is mostly manufactured rather than inherited. It comes from the accumulation of improvements, from interdependence between parts, and from emergence, where many interacting elements produce results nobody designed. All three are consequences of ordinary competent management rather than of any failure.

That is the uncomfortable part, and it is also the hopeful part, since anything you built you can also unbuild.

The accumulation of improvements

Improving a product, a service or a system makes it more complex. Every added feature, exception and refinement increases the number of interactions inside it. Complexity is therefore the principal danger facing any organisation whose innovation program leans heavily on continuous improvement, since the weight arrives one sensible decision at a time.

Think of the software you use daily, the tax code or employment regulation. Nobody set out to build something unusable, and each addition was justified on the day it was made.

The strategic exposure is what interests me. An organisation that keeps adding leaves an opening for a competitor offering something simpler and cheaper, which is the mechanism behind most disruption stories.

Software pricing shows the pattern clearly, with subscription suites accumulating capability and price while single purchase alternatives take ground back by offering less on purpose.

Interdependence, which is different from dependence

Dependence means one thing needs another. Interdependence means each needs the other, so a change in one element changes all of them. More volatility means a system changes faster, which makes it more complex and less predictable, and the loop runs in both directions.

Interdependence takes several forms inside a business. It can be sequential, where each stage feeds the next, pooled, where separate units draw on a shared resource, or reciprocal, where each part continuously adjusts to the others.

Reciprocal interdependence is where the surprises live, and it is also where efficiency programs do the most damage, because removing slack from a reciprocal system removes its ability to absorb variation.

Efficiency programs do most of their damage here, and they do it invisibly. Removing slack from a reciprocal system strips out the capacity that was quietly absorbing variation, so the operation looks leaner on paper and gives way at the first shock.

Emergence, or constant change with no author

Emergence is what happens when many elements interact and produce a result nobody planned or predicted. It is the reason a well-built plan can be obsolete before it is executed, since the environment the plan addressed has reorganised itself while the plan was being approved.

Military accounts of recent conflicts describe exactly this, where the time taken to build and approve a plan exceeded the time the situation stayed still. That observation is the origin of the decision rights argument running through this whole series.

A simple system also tends towards complexity on its own, given time. Someone takes up running in the shoes they already own, then buys proper shoes, then sees a podiatrist, then buys a watch, then changes their diet, then enters a race.

Nobody decided to build that system. It assembled itself from a sequence of individually reasonable choices, which is precisely how it happens inside organisations.

The sheer volume of available information

The volume of information now available is itself a source of complexity, and it is the one that links complexity back to volatility. More sources mean more interactions between claims, more contradictions to reconcile and more effort spent deciding what to attend to. Abundance creates the problem that scarcity used to.

I felt this writing this very series. Which examples to use, which level of detail to pitch at, which material to hold back for the room and which to publish. Every one of those is an interaction rather than a fact, and interactions are what make a task complex.

The organisational version is worse, because the filtering usually happens implicitly. Nobody decides what the business will pay attention to, so attention flows to whatever is loudest, most recent or most easily measured.

That is why I treat attention allocation as a management decision rather than as a personal productivity matter. An organisation that has never named what it will deliberately ignore is carrying every input at once.

The practical version takes ten minutes in a leadership meeting. Name the three signals you will track closely this quarter, name the ones you will check occasionally, and state plainly that everything else is being left alone until something changes.

The complexity gap between the executive suite and the front line

Ask a senior leader what makes their organisation complex and you hear about regulation, competition, international exposure and technology. Ask the people serving customers or making the product and you hear about obsolete processes, unclear responsibilities and instructions that contradict each other. That gap is the single most useful measurement available in this whole topic.

It is also the observation I return to most often in engagements, because it reverses where the work needs to happen.

Two answers to the same question

The executive answer describes complexity arriving from outside, which is real and largely beyond anyone’s control. The front line answer describes complexity generated inside, by design decisions somebody made. Only the second kind can be reduced this quarter, and it is the kind that rarely reaches an executive agenda.

The difference is not a matter of one group being better informed. Leaders look outward and forward, which is their job, while the people doing the work look at the obstacles between them and finishing it, which is theirs.

The consequence is that a complexity program designed at the top addresses the environment, while the complexity your staff experience is untouched. Everyone then wonders why engagement did not move.

I put it bluntly to boards. The complexity of the world as perceived by a chief executive has almost no bearing on the daily work of an operational manager, and mistaking one for the other is expensive in both money and goodwill.

Role clarity is now a regulated duty in Australia

Low role clarity is a named psychosocial hazard under Australian work health and safety law, alongside high job demands, low job control and poor support. Unclear responsibilities and conflicting expectations are therefore a risk an employer must identify and control, rather than an unfortunate side effect of a busy year.

This matters for the argument here because internally generated complexity produces exactly those conditions. Contradictory instructions, unclear ownership and processes that no longer match the work are the everyday form the hazard takes.

Safe Work Australia’s guidance requires businesses to eliminate or minimise psychosocial risks so far as is reasonably practicable, using the same risk management process applied to any other hazard.

The guidance also stresses that these hazards compound. Unclear roles alone may be survivable, and unclear roles combined with heavy demands and little authority produce a risk considerably greater than the sum of the parts.

How to measure the gap in one afternoon

Measuring the complexity gap takes one question asked of two groups. Put the same open question to your executive team and to a sample of frontline staff, collect the answers separately, and compare the categories that appear. The distance between the two lists is your working agenda.

Four prompts get you a usable picture without a survey platform.

1. What is the most complicated part of doing your job well, and where does that complication come from?
2. Which approval or handover slows you down most, and what happens while you wait?
3. Where do you receive instructions that contradict each other, and how do you currently resolve them?
4. Which process do you work around rather than through, and what would break if it were removed?

The fourth question is the productive one. A process people work around has already been judged unnecessary by the organisation, informally and without permission, and removing it is usually the cheapest complexity reduction available.

Do this before you commission anything larger. I have never run this exercise and found the executive picture and the frontline picture in agreement, and the surprise in the room is worth the afternoon on its own.

Why simplification fails and collective seeing works

The answer to complexity is collaboration rather than simplification. Complexity exceeds what any individual can hold, so the useful move is to increase the number of perspectives applied to it. Simplifying a genuinely complex system hides part of it, which delays the surprise instead of preventing it.

There is a real place for removing complexity, and it needs stating precisely so this does not read as an argument for keeping everything.

Managing complexity without simplifying it away

Simplification works on complexity that creates no value, meaning the accumulated processes, approvals, product variants and reports that exist because nobody removed them. It fails on complexity that is inherent to the system, because that kind does not disappear when you stop describing it.

So the goal is never the simplest possible organisation. The goal is the removal of complexity that makes work harder while producing nothing anyone values.

Customers are pushing in the same direction, since products that demand too much of them are increasingly rejected in favour of offers that do less and do it clearly. Fewer and simpler is a defensible commercial strategy as well as an internal one.

The trap is that removal collides with short term value maximisation, because the thing you remove had a sponsor, a budget line and a number attached to it.

Collective intelligence, and what it is not

Collective intelligence is the capacity of a group to reach a solution that none of its members could have reached individually, through interaction and cross-fertilisation. It requires shared information, agreed rules, a process to keep the group on track, and a collective benefit that makes participation worthwhile.

The label gets applied loosely, and the distinction matters. Cooperation and participation run downwards, since a manager asks for an opinion or an idea and then decides.

Collaboration runs sideways. It comes from the interactions between members of a group and their capacity to build on each other’s thinking, and sometimes the manager asked for nothing at all.

This is why calling something collaborative because an opinion was requested is a category error. Under complexity, the difference decides whether the group can see the system or merely comment on the part of it the manager already noticed.

The convening right, and why it is usually missing

The convening right is the authority to pull together the people needed to see a problem whole, without seeking permission first. In most organisations, assembling three people from three functions requires someone senior enough to compel three diaries, which means complex problems queue behind the availability of a sponsor.

This is the cheapest of the four decision rights to move, and the one I raise first with executive teams. Nothing is being spent and nothing is being committed, so the risk is close to zero.

The objection is always about time rather than authority. Which is worth examining, because an organisation that cannot spare four hours to understand a problem will spend four months living with it.

Recognition works the same way. Transferring the power to acknowledge good work from managers to colleagues distributes a judgement that no single person is well placed to make, and it produces a map of who the organisation actually relies on.

What holds a group together long enough to see

Collaboration under complexity requires cohesion, since a group only builds on each other’s thinking when its members want to be in the room. Giving people the tools to work together matters, and giving them a reason to want to matters at least as much. Cohesion is the precondition rather than the outcome.

Four things build it, in my experience of running these groups.

• Proximity, meaning enough contact between people for trust to form before it is needed.
• Equity of attention and responsibility, so that every member contributes time, ideas and expertise rather than a nominated few carrying the group.
• A convergence of interest between what the group achieves and what each member gets from it.
• A shared understanding of the situation, which is why complexity itself has to be explained to everyone rather than held at the top.

The fourth point is the one organisations skip. Teams are asked to work through complex problems without ever being told that the problem is complex, which leaves capable people assuming they are personally failing at something that should be straightforward.

Explain the terrain before you ask anyone to cross it. A team that understands why the usual approach will not work stops looking for the missing answer and starts looking for the missing perspective.

Simplify, collaborate or redesign roles, which investment fits your organisation

Three investments get proposed when complexity becomes a board topic, and they solve different problems. Simplification removes value-free process, collaboration increases the number of perspectives applied to a problem, and role redesign clarifies who owns what. Choose against the symptom you can evidence rather than against the one that is easiest to describe upwards.

The comparison below is written to be used by someone deciding where a budget goes.

Comparing the three investments on explicit criteria

Read across from the symptom column. Notice that the maturity required rises sharply for the third option, because clarifying ownership means naming who no longer owns something, and that conversation is where most complexity programs quietly stop.

Investment	What it changes	Symptom it fits	Managerial maturity required	What it leaves unfixed
Simplification of process and offer	Removes steps, variants, approvals and reports that create no value	Work is slowed by process nobody can justify when asked	Medium, needs someone empowered to delete things	Inherent complexity, which returns as a surprise later
Collaboration and collective intelligence	Increases the number of perspectives applied to a problem	Problems keep being solved in one function and failing in another	Medium to high, needs genuine sideways working rather than consultation	Nothing about ownership, so good analysis still has no home
Role and decision rights redesign	Clarifies who owns what and who may decide without asking	Contradictory instructions, unclear ownership and constant escalation	High, since it requires naming who stops owning something	The external complexity, which is real and stays exactly as it was

One caution about the first row. Simplification is the easiest of the three to sell and the easiest to overshoot, and an organisation that simplifies a genuinely complex system has usually removed the slack that was absorbing variation.

The signals that tell you which one to fund first

Run the complexity gap exercise first, then read the answers. If frontline staff describe process, fund simplification. If they describe not knowing who decides, fund role and decision rights work. And if they describe having to solve problems without access to the people who hold the other half of the picture, fund collaboration.

A fourth signal decides the sequencing. Ask when your organisation last removed a process rather than added one, and if nobody can name an example, start with simplification regardless of what the exercise says.

An organisation with no demonstrated capacity to remove anything will absorb the other two investments as additions. New forums, new roles and new rituals will sit on top of everything already there.

Prove you can subtract once, on something small and visible, before you commit to anything larger.

How I can help you manage VUCA complexity

I work with executive teams, boards and manager communities on complexity, in English and in French. Each of the formats below is built around the same principle, which is that complexity is a group sport, so the work happens with the people who hold the different pieces rather than in a briefing to the people at the top.

Keynotes on leading complex organisations

I speak on VUCA complexity at conferences, leadership offsites and industry summits. The talk works best when it closes with the room writing down their own two lists, the complexity they see from where they sit and the complexity their teams described last week.

Sessions run from a short keynote to a half day with facilitated work between segments, and details sit on my page on managing in a VUCA world.

Workshops on collective intelligence and premortem

Over a half day or two days we run the complexity gap exercise across two groups, work the results through a mixed team that includes the awkward questioners, and project the failure of your current plan before it happens. Teams leave with a list of what to remove and an owner against each item.

Cross-functional range is what makes these groups work, which is the argument behind the T-shaped expert generalist.

Diagnostics and manager communities

For organisations already carrying too much, I run a short diagnostic on where internally generated complexity is concentrated, and on the gap between what leaders and frontline staff describe. The output is a written picture with an owner and a removal date against each item.

I then support the manager community through the removal itself, which is harder than the analysis. Seeing early enough to act helps here, which is the thread running through my work on curiosity in management.

Conclusion: managing complexity is answered together or not at all

Complexity is the letter that most directly contradicts how management was taught. Centralised authority and divided labour assume that someone at the top can hold the whole picture, and under complexity nobody can.

So the honest position for a manager is to stop pretending. Admitting that you do not know is the most reliable way to avoid missing the piece of information that mattered, and pretending otherwise is now the fastest route to losing a team’s respect.

Which leaves two practical moves. Measure the complexity your own people carry rather than the complexity you see from where you sit, and move the convening right down so that a problem can be looked at whole without a sponsor.

Then remove what creates no value, and leave the rest alone. Complexity that produces something worth having is a competitive asset, and complexity that produces nothing is simply weight.

The more people you involve in a problem, the smaller that problem becomes, which is the whole of what twenty years of doing this work has taught me about managing complexity.

Frequently asked questions about managing complexity