How do I lead through ambiguity and uncertainty in a matrix?

Ambiguity in a matrix is not a temporary condition that resolves itself with enough analysis. It is a structural feature of the environment. Most matrix leaders are managing across functions, stakeholders, and competing priorities in situations where nobody, including the people above them, has a clear answer. Waiting for clarity is often not an option. Creating enough direction to keep things moving, without pretending to certainty you do not have, is the actual work.

I was working with a head of operations for a European OEM equipment company through a period of significant market disruption. The business model was under pressure from three directions simultaneously: a cost challenge, a competitive threat from a new category of products, and a regulatory change with uncertain timelines and uncertain implications. The board had agreed that the organisation needed to restructure, but had not agreed what the restructured organisation should look like, because that depended on strategic choices that had not yet been made, because those choices depended on market developments that had not yet fully resolved.

He was responsible for keeping four cross-functional workstreams moving while this was happening. Every workstream lead was asking the same question: what is the direction? He did not have an answer, because the direction had not been decided. His instinct was to get more clarity before communicating. His stakeholders’ experience was that he was being evasive, or that the leadership team was more confused than it wanted to admit. Both conclusions were damaging the programme’s momentum.

The problem was not that he lacked the information. The problem was that he was treating the situation as though it were a technical problem with a correct answer that had simply not been found yet, rather than as an adaptive challenge that would only become clear through action, iteration, and the willingness to commit to provisional directions rather than waiting for definitive ones.

Leading through ambiguity is not the same as tolerating it

Consider two matrix leaders managing similarly ambiguous situations in a post-acquisition integration, where the combined organisation’s operating model has not yet been decided and both legacy organisations are waiting to understand what the future looks like.

The first leader holds back. She does not want to communicate direction that might turn out to be wrong, and she does not want to commit resources to an approach that might need to be reversed. She is thorough, careful, and well-intentioned. What her team experiences is a vacuum. People who cannot see direction from their leader generate their own, usually in the most anxious direction available. The integration stalls, not because nobody knows what to do, but because everyone is waiting for someone else to decide.

The second leader makes provisional commitments. She is clear about what she does not know and equally clear about what she is treating as directional enough to act on: a set of working assumptions that are explicitly labelled as provisional, that give enough shape for each workstream to move forward, and that are explicitly open to revision as new information arrives. She does not pretend to certainty she does not have. She also does not use the absence of certainty as a reason to remain still.

Same ambiguity. Same absence of definitive answers. The difference is not access to information. It is the understanding that in a complex environment, action creates the data that makes better decisions possible, and that waiting for clarity before acting is a reliable way to produce the thing you are most trying to avoid: a situation where time has run out and the decision is made for you by events.

This distinction is the most important thing to understand about leading through ambiguity. Most leadership situations are either obvious, where the right response is clear and the skill is execution; or complicated, where the right response requires expertise and careful analysis but can in principle be found; or complex, where the right response cannot be known in advance and only becomes visible after the organisation has acted and observed what happens. The problems that are most exhausting and most damaging in a matrix are the complex ones, and the most common cause of failure in complex situations is applying the tools that work well in complicated ones: more analysis, more planning, more expert input, more time. These tools are not wrong. They are addressing the wrong kind of problem.

This article looks at how to lead through ambiguity and uncertainty as an advanced capability in matrix leadership. It is one of the final articles in this series before the synthesis piece that closes it. The foundation article, How do successful leaders create commitment in matrix organisations?, introduced the full framework. The series covers what a matrix organisation is and why companies adopt one, how to build influence without authority, how to build a powerful internal network, how to convene people who do not report to you, how to create accountability without authority, how decisions get made in a matrix organisation, how to manage competing priorities in a matrix, how to resolve conflict between departments, and how to build trust across organisational boundaries.

Why ambiguity is particularly difficult to lead through in a matrix

Assuming someone above you has the answer: In a hierarchy, uncertainty at your level is often reducible by escalation: your manager has a clearer picture than you do, and getting access to that picture is part of what managing upward means. In a matrix, this relief is often unavailable. The ambiguity is frequently structural. The organisation does not know the answer. The executive team does not know the answer. The board has not decided. Treating this as a temporary state that will resolve itself once someone makes a decision is a reliable way to run out of time without having moved.

Applying technical solutions to adaptive challenges:  Many of the most difficult situations in matrix leadership are not complicated: they are complex. A complicated problem has a right answer that can be found with enough expertise and analysis. A complex challenge requires new learning, new behaviour, and adjustment from multiple parts of the organisation, and the right answer is not available until people have begun to act and have observed what happens. Applying analysis and planning to an adaptive challenge, however rigorously, will not produce clarity. It will produce a better-informed stall.

Treating the absence of certainty as a reason to stay still: The most common individual response to genuine uncertainty is to gather more information before committing. This is a sensible instinct in stable, complicated environments. In a complex environment, it is often counterproductive, because the information that would reduce uncertainty only becomes available after action has been taken. The leader who waits for clarity before moving is waiting for something that will not arrive until they have moved.

Filling the vacuum with the wrong kind of certainty: The opposite error is to project false certainty: to communicate direction with more confidence than is warranted, because the discomfort of the team’s uncertainty feels like a problem to be solved. This works in the short term and causes serious damage in the medium term, because when the false certainty turns out to be wrong, the trust damage is compounded by the knowledge that the leader knew they did not know and chose to conceal it.

Getting trapped in a single story: In conditions of uncertainty, leaders are particularly susceptible to attaching to a single interpretation of what is happening and defending it rather than holding multiple interpretations simultaneously and remaining genuinely open to the one that turns out to fit best. Research on leadership in complexity identifies this as one of the most common and most damaging instincts: the pull towards a simple, coherent narrative in situations that are actually ambiguous, because the simple narrative feels safer and more controllable than an honest acknowledgement of genuine uncertainty.

Confusing clarity of process with clarity of outcome: A structured process, a clear roadmap, a well-designed governance framework, can give a matrix programme the appearance of direction without actually providing it. Leaders who spend significant energy designing and communicating the process often do so because the outcome is unclear, and the process feels like a substitute for the direction that is not there. It is not. People can see through a well-formatted plan that nobody knows where it leads.

Six practices for leading through ambiguity and uncertainty in a matrix

Diagnose which domain you are actually in before deciding how to respond

The most costly mistake in genuinely ambiguous situations is not making the wrong call. It is applying the right tool to the wrong kind of problem. A framework developed by researchers studying complex systems offers a practical diagnostic that most matrix leaders find immediately useful: the idea that situations fall into distinct domains, each of which requires a different leadership response.

In an obvious situation, cause and effect are clear, the right response is well known, and applying established best practice is the appropriate move. In a complicated situation, cause and effect are discoverable but not immediately visible: expertise and careful analysis are needed, but in principle the right answer can be found. In a complex situation, cause and effect are not knowable in advance: the right response only becomes visible after the organisation acts and observes what happens. In a chaotic situation, no clear relationship between cause and effect exists at all, and the first priority is to act quickly to establish enough stability to begin making sense of what is happening.

The reason this diagnostic matters for a matrix leader is that most of the situations that are most exhausting and most damaging in a matrix are complex, and most of the instinctive responses to them are appropriate for complicated situations. Commissioning more analysis, bringing in more expertise, designing a more rigorous process: these are exactly the right moves in a complicated situation and consistently insufficient in a complex one. A situation that cannot be resolved by more analysis is not necessarily harder than one that can. It is different in kind, and it requires a different response: not analyse-design-implement but probe-sense-respond. Small, contained actions that generate information, followed by observation of what those actions produce, followed by adjustment and further action.

There is also a specific warning worth naming. Situations that look obvious, where cause and effect seem clear and the right response feels settled, can tip very quickly into chaotic if the underlying assumption that made them feel obvious collapses. Matrix programmes that have been running smoothly on a set of stable assumptions are particularly vulnerable at the point when those assumptions change: the apparent obviousness of the situation can mask the fact that complexity has been quietly accumulating beneath the surface. The cliff edge between what feels settled and what suddenly becomes turbulent is one of the most important features of genuinely complex environments, and it is almost always invisible from inside the situation until after the fall.

A strategy director I worked with spent six months and significant resource on an analytical approach to a market entry question. The analysis was thorough. Every time it produced a provisional answer, new data emerged that complicated it. When she reframed the question using this diagnostic, she recognised immediately that she had been treating a complex situation as a complicated one. The answer was not in the analysis. It was in the market’s response to a specific action. She designed three targeted probes, each testing a different assumption, and within six weeks had more useful information than a year of analysis had produced.

The head of operations found the same reframe useful for the restructuring programme. Three of the four workstreams had a clear enough technical shape: they needed decisions about systems, reporting lines, and resource allocation. Those were complicated rather than complex: analysable, plannable, resolvable with expertise. The fourth workstream, which was supposed to be about capability development but kept stalling, turned out to be genuinely complex: it required the functional leads to agree on what good leadership looked like in the new organisation, and they had different answers they had not yet surfaced. No amount of analytical rigour was going to move it until the right actions had been taken to make those differences visible.

Questions you can use to diagnose which domain you are actually in:

• If you brought in the best available expertise on this question, would they be able to give you the answer? Or would the answer still depend on what the people involved choose to do and what the environment produces in response?
• Has this problem been solved before, in this organisation or others, and has the solution held? If the same problem keeps returning in different forms, it is probably complex rather than complicated.
• Does making progress require people to change something about how they work or what they value, rather than simply acquiring new information or applying existing capability?
• What would you need to observe, in response to a specific action, to know whether you are moving in the right direction? If you cannot answer that question, the situation is probably complex.
• Is there anything about the current situation that looks obviously settled but that you have not tested recently against actual evidence?

Practical actions:

• Before committing further resource to an analytical approach that has not produced a workable conclusion, ask explicitly whether the situation is complicated or complex. If you cannot identify in advance what information would resolve it, it is probably complex.
• For complex situations, replace analyse-design-implement with probe-sense-respond: design the smallest action that would generate meaningful information, take it, observe the result, and adjust before acting again.
• For situations that have been feeling obvious and settled, run a brief test: what assumption is most central to the current approach, and when did you last check that it was still holding? Treat an assumption that has not been tested in three months as a candidate for attention.
• Name the domain to your programme team and your key stakeholders. A team that understands it is working in a complex rather than a complicated environment will approach the work differently and will be less likely to interpret the absence of a clear answer as a failure of leadership.

Act to learn rather than wait to know

In a complex environment, the information required to make a good decision is often not available until after a decision has been made and its consequences observed. This is not a failure of analysis or foresight. It is a structural feature of complex systems. The right response is not to analyse more, plan more, or wait longer. It is to act in ways that generate the information that better decisions require, while keeping the actions small enough to be reversible if the information they generate points in a different direction.

This is a fundamentally different relationship with action than most leaders are trained to have. The conventional model, in which you gather information, make a decision, and then act, assumes that the information exists and is accessible before action. In complex environments, the information is frequently created by the action. We make sense of what has already happened, not of what is about to happen. Action produces the data that sensemaking requires.

The practical discipline has three specific components, each of which is important and each of which is easy to shortcut. First, keep experiments small in scope and cost, ensuring that failure does not harm the wider system. A probe that is too large to fail safely is no longer a learning experiment: it is a commitment that has borrowed the language of experimentation. Second, set clear hypotheses before acting: define in advance what you expect to see, how you will measure it, and what would count as evidence that the direction is viable or not. An experiment without a hypothesis generates activity rather than learning. Third, build short feedback loops that enable rapid adjustment rather than waiting until the end of a project cycle. The value of experimentation comes from the speed at which it generates information, and a long cycle time eliminates much of that value.

There is also a dimension worth naming that most descriptions of experimentation in business omit: the impact of your experiments on the people and systems around you. A lateral leader who runs experiments across functions without considering how those experiments affect the functions involved is generating learning for themselves while generating disruption for others. The ethical and relational discipline of experimentation in a matrix context requires asking, before each experiment, who will be affected by it and how, and involving those people early enough that the experiment is genuinely collaborative rather than something done to them in the name of organisational learning.

A business development director at an international professional services firm described a market positioning question that had been under consideration for nearly a year. She had commissioned research, run stakeholder workshops, and reviewed the competitive landscape thoroughly. None of it had resolved the question. She changed her approach: she identified three small client conversations that would generate direct evidence about how two different positioning options were received in practice, defined clearly what she would conclude from each possible response, and treated those conversations as experiments rather than as sales meetings. Within six weeks she had more useful information than a year of analysis had produced.

The head of operations in the OEM equipment business used the same approach with a workstream decision that had been stalling. Rather than continuing to seek agreement on the principle of the new operating model, he proposed running three functions on a provisional version of it for eight weeks and reviewing what they had learned. He involved each function’s lead in designing the experiment, specifically so that the experiment belonged to all of them rather than being something he had designed and imposed. The workstream leads who had been unable to agree on the principle were able to agree on the experiment, because it was reversible, had a clear review point, and had been genuinely co-created. The experiment produced specific, observable evidence that changed the nature of the conversation considerably.

Questions you can use to find the smallest useful action in an ambiguous situation:

• What is the smallest action that would produce meaningful information about which direction to move in, without committing resources or positions that would be expensive to reverse?
• Is there a way to test the most critical assumption in the current plan without running the full plan?
• What would you need to observe, in response to a specific action, to conclude that the direction you are considering is viable or not?
• Who will be affected by this experiment, and have you involved them in designing it rather than simply informing them that it is happening?
• What has the programme already tried that has not worked as expected, and what has that told you that analysis alone could not?

Practical actions:

• In any situation where analysis has run for more than two months without producing a workable conclusion, ask whether there is a small action that would generate more useful information than more analysis would.
• Before running any experiment, write down the hypothesis: what do you expect to see, and what would you conclude if the result is different from what you expected? An experiment without a written hypothesis is harder to learn from.
• Keep experiments reversible wherever possible, and confirm explicitly with the people involved that the experiment can be stopped or changed if the evidence points in a different direction.
• After any significant experiment, evaluate not just the outcomes but the side effects: who was affected by it, how, and what that implies for how the next experiment is designed and with whom.

Make your provisional interpretation explicit and shareable

In ambiguous situations, people fill the information vacuum created by leadership silence with their own interpretations. In a matrix, where different functions and stakeholders are each filling the vacuum independently, those interpretations quickly diverge, and the divergence compounds the ambiguity rather than reducing it. The leader who waits for certainty before communicating is not protecting their credibility. They are creating a situation in which multiple incompatible interpretations of the same uncertainty are shaping behaviour across the organisation simultaneously.

The alternative is not false certainty. It is disciplined uncertainty: a clear, specific articulation of what is known, what is not known, and what is being treated as directional enough to act on despite the uncertainty. This is different from vague reassurance and different from false confidence. It is an honest account of where things actually stand, with enough provisional shape to give the organisation something to work with while the ambiguity resolves.

Meaning is not found in complex situations, it is actively made. The leader who articulates a provisional interpretation, even if it is explicitly incomplete and subject to revision, is doing something more valuable than the leader who remains silent while waiting for the right answer. They are creating a shared reference point that makes coordinated action possible, building trust through honesty about what is not yet known, and modelling the kind of disciplined uncertainty that complex environments require of everyone.

The specific discipline here has two parts that most leaders collapse into one. The first is building the framework: translating the raw data and observations you have gathered into a working interpretation that organises complexity into something intelligible. The second, which is equally important and more often skipped, is sharing that framework widely enough to receive the critique and the additional data that will make it more accurate. A map that is shared only with the people who helped build it is not a shared map. It is a private interpretation that has been communicated. A genuinely shared map is one that has been exposed to the people most affected by it, with an explicit invitation for them to identify what it is missing and where its assumptions are wrong.

This second step requires a specific kind of humility: the willingness to present a working framework not as a finished product but as a draft. Leaders who present their maps as conclusions invite people to accept or reject them. Leaders who present them as working hypotheses invite people to improve them, which produces both a better map and a broader sense of ownership in the people who contributed to it.

A programme director leading a post-merger integration described losing the confidence of her steering group not because she lacked answers but because her communication pattern had created the impression that she had answers she was withholding. Every steering group update had been carefully worded to avoid committing to positions that might turn out to be wrong. The steering group concluded that she was managing them rather than informing them. When she changed her approach, opening the next steering group with a direct account of what was clear, what was unclear, and what she was treating as provisionally directional and why, several members said afterwards that it was the most useful update they had received in the programme, not because it contained more information but because it contained more honesty.

The head of operations described a significant shift in his working relationship with the four workstream leads when he changed how he opened their weekly calls. He had been opening with a structured agenda that implicitly suggested he had more clarity than he did. When he began opening with a three-part framing, here is what I know, here is what I do not know, here is what I am treating as direction for this week pending a clearer picture, the character of the conversations changed. People stopped asking him to confirm a certainty he did not have and started contributing to the shared task of working through the uncertainty together.

Questions you can use to develop and communicate a provisional interpretation:

• What do you actually know with reasonable confidence, and what are you assuming or hoping is true?
• What would you need to see in the next four weeks for the current direction to hold, and what would you need to see for it to change?
• Is there a way to describe your current best interpretation of the situation that is honest about its provisional nature without abandoning the directional commitment that the organisation needs to function?
• When you shared this interpretation most recently, did you present it as a draft to be improved or as a conclusion to be accepted? What happened as a result?
• Who has not yet had the opportunity to critique your current map of the situation, and whose critique would most likely surface something you have not yet seen?

Practical actions:

• In your next significant stakeholder communication, structure it explicitly around three categories: what is known, what is not yet known, and what is being treated as provisionally directional and why.
• When you catch yourself communicating in a way that implies more certainty than you have, stop and restate. The short-term cost of admitting uncertainty is consistently lower than the medium-term cost of being seen to have concealed it.
• Share your current working interpretation with at least one person who is most likely to disagree with it or to have access to data you do not, before you share it more broadly. Treat what they tell you as the most valuable input available.
• After any significant update to your working interpretation, tell the people who contributed to the change what changed and why. Visible integration of feedback is what keeps people engaged in the shared sensemaking process rather than withdrawing from it.

Regulate the heat: keep uncertainty productive without letting it become paralysing

Acknowledged uncertainty has a temperature. Too cold, and people assume the ambiguity is being managed by someone above them and disengage from the adaptive work the situation requires. Too hot, and people become anxious, defensive, and unable to think productively about the questions they most need to be thinking about. The lateral leader’s job in genuinely ambiguous situations is to keep the temperature in the productive zone: warm enough that people are genuinely engaged with the uncertainty, cool enough that they can function well in the presence of it.

This is a calibration that has to be done differently for different stakeholders. The same level of acknowledged uncertainty that is productive for a senior leader with high tolerance for ambiguity can be paralysing for an operational team that needs enough stability to plan next week’s work. And the same level of reassurance that keeps an operational team functional can generate complacency in a senior leadership group that needs to be genuinely grappling with the strategic questions.

A matrix lead at a European infrastructure business described managing two very different stakeholder groups simultaneously during a period of significant regulatory uncertainty. The executive team needed to be held in genuine engagement with the uncertainty: if they were too reassured, they stopped making the exploratory decisions that the situation required. The project delivery teams needed enough provisional direction to plan their work: if they were held in the same level of acknowledged uncertainty as the executive team, they stopped functioning. He managed the temperature differently in each direction, not by being inconsistent about what he knew, but by being thoughtful about what each group needed to be able to hear and act on.

The head of operations in the OEM equipment business made the same distinction across his four workstream leads. Two of them had the seniority and the temperament to work productively with a high level of acknowledged uncertainty. Two needed more provisional direction than he might ideally have provided at the strategic level, because without it their teams stopped moving. He gave each group what it needed to remain functional, while being honest with all of them about what was provisional and why.

Questions you can use to calibrate the heat of acknowledged uncertainty for different stakeholders:

• For each of your key stakeholders, what level of acknowledged uncertainty is productive and what level becomes paralysing?
• Are there stakeholders you have been keeping at too cool a temperature, where the reassurance you are providing is creating complacency rather than engagement?
• Are there stakeholders you have been holding at too hot a temperature, where the acknowledged uncertainty is generating anxiety that is reducing rather than improving their contribution?
• What does each group need to remain functional, and is that different from what would be ideal if the uncertainty were not a constraint?
• Is the way you are communicating about the uncertainty helping people stay engaged with the real questions, or is it creating a static they cannot work through?

Practical actions:

• Map your key stakeholders against their capacity for acknowledged uncertainty, and calibrate what you communicate to each group accordingly. This is not deception. It is thoughtful communication that gives each group what it needs to remain functional and engaged.
• For stakeholders who need more provisional direction than you would ideally provide, make the provisionality explicit: I am giving you this as a working direction rather than a confirmed one, so that your team can keep moving, and here is when we will review it.
• When a group is displaying anxiety that is reducing their functional capacity, ask directly what would help them work through it rather than around it. Sometimes naming the uncertainty as shared rather than as something they need to wait for you to resolve is enough to reduce the temperature.
• Review the heat calibration at each significant milestone. What was the right temperature for month three of the programme may not be the right temperature for month nine.

Get to the balcony: create distance from the immediate to see the pattern

In a matrix, the demands of stakeholder management are relentless. There is always a call to take, a concern to address, a conflict to facilitate, a commitment to track. The accumulation of immediate demands can make it genuinely difficult to step back far enough to see whether the pattern of what is happening is what it appears to be from inside the action, or something different.

The capacity to step back from immediate demands and observe the larger pattern is one of the most consistently underinvested capabilities in matrix leadership, and one of the most consequential. A leader who is always engaged with the immediate situation will find themselves addressing the symptoms of a pattern they have never had enough distance to see. The strategic moves, the decisions about what to pay attention to and what to let go, the recognition that the problem has changed shape while the programme has been addressing its earlier version: these are all only visible from a distance that immediate demands routinely prevent.

The discipline of getting to the balcony has four specific components that are each easier to describe than to sustain. The first is deliberate scanning across a genuinely wide range of data sources, not just the financial reports and KPI dashboards that arrive automatically, but qualitative signals, weak indicators, information from adjacent domains, and perspectives from people the programme does not typically consult. The signals that matter most in complex environments are often not the loudest ones. They are the anomalies, the customer complaints, the unusual patterns in staff feedback that get dismissed as noise before anyone has asked whether they are pointing at something real.

The second is actively involving others to stretch your mental models. The cross-boundary position of a lateral leader in a matrix is one of its most underused assets for pattern recognition. A leader who is genuinely spanning multiple functions has access to perspectives that no single function possesses. The discipline is to use that access deliberately: to bring together people whose different vantage points will surface things that any single perspective would miss, and to create the conditions for those perspectives to genuinely challenge each other rather than simply coexisting.

The third is seeing with new eyes: suspending the labels and categories that make the current situation feel legible, long enough to ask whether the legibility is real or whether the organisation is simply applying familiar patterns to a situation that has genuinely changed. The music industry’s failure to see digital distribution as an opportunity rather than a threat is one of the most cited examples of this trap. The more familiar version in a matrix context is the programme that continues to treat a stakeholder as a supporter because they were one six months ago, without noticing that their priorities have shifted.

The fourth is connecting directly to the frontlines: the operational staff, the customers, the people who encounter the reality of the situation first and whose observations are often the earliest and most accurate signal of what is actually happening. In a matrix, this requires deliberate effort, because the lateral leader’s natural position is at the intersection of functions rather than at any single function’s operational core.

A regional director leading a market development programme described a pattern shift that had been visible, in retrospect, for three months before he noticed it. A set of signals, each individually explicable as ordinary programme noise, had together been indicating that a key assumption in the programme logic was no longer holding. He had been too close to the immediate work to see the signals as a pattern rather than as isolated events.

The head of operations built a specific practice into his work for the restructuring programme: every Friday afternoon, he blocked two hours for a single question. What pattern am I not seeing from inside the programme that I would see if I could look at it from outside it? He described the practice as initially uncomfortable and subsequently indispensable. Several of the most useful course corrections he made in the programme originated in that Friday afternoon question rather than in any meeting or stakeholder conversation.

Questions you can use to get to the balcony and stay there:

• If someone who knew the context but had no stake in the current approach looked at what the programme had been doing for the last six weeks, what pattern would they see that you are too close to see yourself?
• What signals has the programme been receiving that have been filed as noise or explained away as isolated events? Is there a pattern in them?
• Is the problem you are currently working on the same problem you started with, or has it changed shape while you have been addressing its earlier version?
• Have you heard recently from the people closest to the operational reality of this programme, not in a formal review meeting but in a direct, unstructured conversation?
• What label or category are you currently applying to this situation, and when did you last test whether it still fits?

Practical actions:

• Protect regular time, weekly if possible, for observation rather than action: time that is not committed to meetings, conversations, or deliverables, and that is used specifically to ask what pattern you are not seeing from inside the work.
• Deliberately widen the data sources you are drawing on beyond the standard programme reporting. Identify at least one weak signal, an anomaly, an unusual complaint, an unexpected result from an experiment, and follow it far enough to understand what it is pointing at.
• Identify one person outside your programme who has enough context to be useful and enough distance to see things you cannot, and schedule a regular conversation with them whose explicit purpose is to help you see the programme from further away.
• At every significant programme milestone, write a one-page account of what the programme has learned, as opposed to what it has delivered. This is more useful than a progress report for identifying what the programme actually needs next.

Hold your interpretation lightly: notice and escape the mindtraps

The most reliable signal that a leader is in a mindtrap in a complex environment is that their current interpretation of the situation feels obviously right. In a stable, complicated environment, a high degree of confidence in a well-reasoned interpretation is appropriate. In a complex environment, that same confidence is often the clearest available warning that the interpretation is too simple for the situation it claims to describe.

Research on leadership in complexity identifies five recurring patterns through which experienced leaders systematically mislead themselves in ambiguous situations. The pull towards a simple, coherent story that reduces the discomfort of genuine uncertainty. The sense of rightness about the current interpretation that makes contradicting data easy to discount. The tendency to seek agreement rather than productive disagreement, because agreement feels like alignment and disagreement feels like a problem to be managed. The fixation with controlling outcomes that are, by the nature of complex environments, not controllable. And the protective move of the ego, which makes admitting that the current approach is not working harder than almost any other leadership behaviour.

Each of these is a normal and functional response in stable environments. Each becomes a trap in complex ones, because the behaviours that help a leader be clear, decisive, and aligned in a straightforward situation actively undermine their ability to navigate a genuinely ambiguous one. The most damaging moment is not when a leader makes a wrong call. It is when they make a wrong call and then defend it against evidence rather than updating it.

The control mindtrap deserves particular attention in a matrix context. The instinct to control outcomes is especially strong for leaders who have programme accountability but no formal authority, because the absence of authority makes the need for control feel even more urgent. In a complex environment, that instinct consistently backfires: attempts to control what cannot be controlled typically produce resistance, workarounds, and a withdrawal of the discretionary cooperation that the lateral leader depends on entirely. The more productive question is not “how do I control this” but “what conditions can I help create that would make the right outcome more likely to emerge.” That shift in question changes what the leader actually does, and it changes the nature of the relationship between the leader and the people they need.

A transformation director at an engineering company described a pattern she recognised in herself during a three-year digital transformation programme. The programme had been designed around a clear theory of change that had made complete sense at the time. Eighteen months in, the market and the technology environment had shifted in ways that made two of the core assumptions in the theory of change questionable. She had the data. She noticed it but had filed it as noise because it conflicted with the programme’s direction and because acknowledging it would have required a significant renegotiation with the steering group. Six months later, the assumptions had collapsed sufficiently that the renegotiation was unavoidable. The delay had cost the programme credibility and six months of resource applied to an approach that no longer fitted the situation.

The head of operations described a version of the simple-story mindtrap in his own work. He had built a clear narrative about why the restructuring was taking the shape it was, a narrative that had been genuinely useful for communicating the direction to stakeholders. At a certain point, the narrative had stopped helping him think and started helping him avoid thinking. When a senior colleague named this directly, he found it immediately recognisable: he had been using the coherence of the story as a reason not to engage with the data that was making the story slightly less coherent week by week.

Questions you can use to notice and escape the mindtraps in complex situations:

• Is the current interpretation of this situation feeling more obviously right than the complexity of the situation warrants? If so, what is the data that the interpretation is having to work hardest to accommodate?
• What would someone who fundamentally disagreed with the current direction see in the situation that you are not currently paying attention to?
• Is there a piece of feedback or a signal from the environment that you have heard but filed rather than acted on, because it conflicted with the current approach?
• Are the conversations in the programme oriented towards agreement and alignment, or towards genuine disagreement that expands the solution space?
• What would have to be true for the current direction to be wrong, and when did you last take that question seriously?

Practical actions:

• Build a standing question into your regular rhythm with your closest programme colleagues: what are we currently most confident about, and is that confidence warranted by the evidence? Treat the answer as a diagnostic for which mindtraps you are currently most at risk of.
• Seek out at least one person per month who will tell you something you do not want to hear about the programme, and create the conditions for them to do so honestly. The candid conversation that is uncomfortable to have is consistently more valuable than the comfortable conversation that confirms the current direction.
• When you notice yourself explaining away contradicting evidence, stop and ask what you would conclude if you took the evidence at face value rather than fitting it into the current story.
• At every significant decision point, run a brief premortem: if this decision turns out to be wrong in twelve months, what is the most likely reason? The question surfaces the assumptions you are most invested in protecting and gives you the opportunity to test them before the decision rather than after it.

Wrapping up

Leading through ambiguity and uncertainty in a matrix is not a capability that can be reduced to a more thorough planning process or a more sophisticated analytical framework. The conditions that make ambiguity difficult in a matrix, structural uncertainty that nobody above you has resolved, multiple stakeholders each needing something slightly different, action required before the answers are clear, are precisely the conditions that make those tools insufficient. The lateral leader who can diagnose which domain they are actually in, act to learn rather than wait to know, make provisional interpretations explicit and shareable, regulate the heat of acknowledged uncertainty for different stakeholders, step back far enough from the immediate to see the pattern, and hold their current interpretation lightly enough to update it when the evidence requires, is not simply more comfortable with uncertainty. They are more effective in it, which in a matrix environment where uncertainty is structural rather than temporary, amounts to the same thing.

Three questions for reflection

1. Think of a significant challenge you are currently managing that has not responded to technical approaches. Is it complicated or complex, and if it is complex, what would the smallest useful probe look like?
2. Is there a provisional interpretation of your current situation that you have been keeping to yourself because it is not yet certain? What would happen if you shared it as a working hypothesis and invited genuine critique?
3. In the last month, what evidence have you received that conflicted with the current direction of your programme, and did you file it, explain it away, or take it seriously?

Inspiration

Ancona, D. (2012) ‘Sensemaking: Framing and acting in the unknown’, in Snook, S.A., Nohria, N.N. and Khurana, R. (eds.) The handbook for teaching leadership: Knowing, doing, and being. Thousand Oaks, CA: Sage, pp. 3–19.

Garvey Berger, J. (2019) Unlocking leadership mindtraps: How to thrive in complexity. Stanford, CA: Stanford University Press.

Heifetz, R.A. and Linsky, M. (2002) Leadership on the line: Staying alive through the dangers of leading. Boston, MA: Harvard Business School Press.

Heifetz, R.A., Linsky, M. and Grashow, A. (2009) The practice of adaptive leadership: Tools and tactics for changing your organization and the world. Boston, MA: Harvard Business Press.

Snowden, D.J. and Boone, M.E. (2007) ‘A leader’s framework for decision making’, Harvard Business Review, November 2007. See my CYNEFIN article.

Weick, K.E. (1995) Sensemaking in organizations. Thousand Oaks, CA: Sage Publications.

Wheatley, M.J. (1992) Leadership and the new science: Learning about organization from an orderly universe. San Francisco: Berrett-Koehler Publishers.

My Complexity Leadership library explores this topic in great depth.