When a company’s culture turns toxic, who gets the call? Almost always, it’s HR. Organisations hand them the problem, give them a budget for workshops and engagement surveys. And expect them to fix what took years to break. But here’s the uncomfortable truth. No HR team, no matter how talented, can build or repair organisational culture on their own.
Organisational culture is not a policy issue. A new set of values pinned to the wall or a Friday afternoon team lunch will not solve it. Leaders at every level of the business shape it every day through their behaviour. What leaders tolerate, reward, ignore, and model becomes the culture, whether they intend it or not.
In this analysis, we will examine why leadership, not HR, holds the real levers of cultural change. We will explore how culture forms, where HR’s role genuinely begins and ends. And what leaders must start doing differently if they want lasting results. If you manage people or influence strategy, this is a conversation you cannot afford to sit out.
Why Most Culture Change Fails Before It Begins
Up to 70% of organisational change efforts fail, and that figure has remained stubbornly consistent across decades of research. McKinsey’s own transformation data tells us that fewer than one in three organisations successfully improve performance. Whilst sustaining those improvements over time. These are not figures from a single bad year or a narrow industry sample. Thousands of well-funded and well-intentioned change programmes produced these outcomes, yet structural flaws doomed most of them from the outset. The reason they fail is not complexity, budget, or timing. The reason they fail is that organisations persistently confuse communication with culture change. The two are not the same thing.
The most common culture change mechanisms are values posters, away-day retreats, and policy documents. They are also the least effective. Each of these interventions operates at the level of the visible. What the organisation says it believes, how it chooses to describe itself, what it writes down and circulates.
None of them touches the behavioural layer, which is where culture actually lives. Culture is not a statement of intent pinned to a breakroom wall.
It is the sum of what behaviour gets rewarded, what gets quietly tolerated. And what gets repeated across thousands of daily interactions between leaders and their teams. When those daily interactions contradict the values on the poster. Employees do not update their understanding of the culture to match the poster. They update it to match the behaviour they observe. Research into employee reactions to organisational change confirms that responses operate across cognitive, emotional, and behavioural dimensions simultaneously.
Interventions that address only the cognitive layer, the messaging and the meaning-making, leave the emotional and behavioural layers entirely untouched.
Organisations design most culture change programmes as communications exercises because they can measure, budget for, and complete them. You can schedule a launch event, produce a values framework. Run a series of town halls, and mark the project as delivered. Behaviour change offers none of that comfort. It is slow, non-linear, and difficult to measure in the short term. Deeply dependent on what leaders actually do rather than what they say. Organisations default to communications because the alternative requires sustained leadership accountability. That is considerably harder to package into a programme plan. Change management data consistently identifies leadership misalignment as a primary driver of transformation failure. Yet most culture programmes are designed in ways that allow senior leaders to sponsor change. Without materially changing their own behaviour.
The delegation problem compounds this further. When senior leaders hand culture to HR, they make a cultural statement before the programme even launches. That culture is a support function rather than a core leadership responsibility. HR can facilitate, measure, and report on culture. HR cannot model it at the level that shifts norms across an organisation. Leaders must do that work through one-to-one conversations. The way they conduct performance reviews, and the behaviours they choose to challenge or excuse.
The failure pattern that follows is entirely predictable. Senior leadership announces a cultural transformation. Middle managers absorb the messaging and relay it downward without conviction. Because they have seen this before and have no reason to believe this time is different. Frontline behaviour remains unchanged because the reward structures, the decision-making norms, and the tolerance for certain conduct have not shifted. The gap between the stated values and the lived experience widens, and what fills that gap is not confusion. It is cynicism. And once cynicism takes hold, leaders find it significantly harder to shift than the original cultural problem they set out to solve.
What Organisational Culture Actually Is, and Why the Definition Matters
Think of organisational culture as the operating system running silently beneath everything your organisation does. It determines how people make decisions when no policy covers the situation. How they behave when leaders are not watching, and what they prioritise when resources run short. Most leaders spend their time working on the visible applications, the strategy decks, the KPIs, the restructuring plans. Without ever examining whether the operating system beneath those efforts is capable of running them. That oversight is precisely why so many change programmes fail before they find their footing.
Gallup’s research on organisational culture frames culture as the pattern of shared assumptions, values, and beliefs that governs how people behave in an organisation. That definition is useful, but it carries a hidden complexity that most leadership teams miss. There are, in practice, two versions of every organisation’s culture.
The first is the espoused culture. The values on the wall, the purpose statements in the annual report, the leadership behaviours listed in the competency framework. The second is the enacted culture. What the organisation’s actual behaviour reveals about what it truly prioritises.
Edgar Schein’s foundational three-layer model captures this precisely. Artefacts are visible and manageable. Organisations state espoused values to express what they aspire to achieve. Underlying assumptions are unconscious and enormously powerful. Most leaders operate at the artefact level and wonder why nothing changes at the assumption level. The gap between what an organisation claims to be and what its behaviour reveals it actually is tends to close only under pressure. Rarely in a flattering direction.
Culture is also not the single, unified thing that most leadership communications treat it as. Research published through the National Institutes of Health draws on Hofstede’s cultural dimensions to show how organisations contain multiple, often competing, sub-systems across departments, seniority levels, and geographies. A sales team operating under a results-oriented culture will develop different norms. Different assumptions and different unspoken rules than a sales team operating under a process-oriented culture. Treating the organisation as a single cultural entity is not just analytically imprecise. It is a strategic error that leads to interventions which work in one part of the business and quietly undermine everything in another.
In 2026, five simultaneous pressures intensify this complexity and reshape organisational culture. AI adoption, the normalisation of hybrid work, skills-based organisation design, generational workforce complexity, and the growing imperative for ethical leadership. Each of these is capable of fracturing a fragile culture independently. Together, they create conditions in which cultures that appeared stable are revealing significant fault lines. AI adoption alone is reshaping how organisations make decisions and whom they trust to make them. Hybrid work has removed the proximity on which so much informal culture depended. Skills-based design is challenging the positional authority that held traditional hierarchies together.
The argument that culture drives performance is not new. What is new is the accountability infrastructure forming around it. Organisations are now measuring cultural health through engagement scores, wellbeing indicators, and behavioural data. Turning what was once a qualitative assertion into something quantifiable and governable. Culture is no longer something leaders can simply declare. They have to demonstrate it, and increasingly, the evidence either supports that claim or contradicts it.
AI Adoption Is Not a Technology Challenge, It Is a Culture Challenge
DDI’s 2026 Global Leadership Forecast identifies human-AI partnerships as one of the top defining trends reshaping leadership. Yet most established culture frameworks have not caught up with this reality. Organisations designed models around values alignment, psychological safety, and communication norms, for a world in which human behaviour represented the most disruptive variable. None of them formally incorporate AI governance as a cultural dimension. Instead, organisations consistently treat AI adoption as a technology procurement decision, a compliance obligation, or an IT implementation project. That framing is precisely why so many AI rollouts stall. Generate resistance or produces results that disappoint relative to the investment. The technology rarely fails. The culture does.
What makes AI adoption distinct from any previous technology shift is the diagnostic clarity it brings. When AI tools arrive in an organisation, they act as an unforgiving mirror, reflecting back the true state of the culture underneath. Psychological safety, which determines whether people feel safe enough to admit confusion, flag errors, or voice scepticism, becomes immediately visible in how employees engage with new AI capabilities.
Trust in leadership transparency surfaces in whether teams believe the stated reasons for AI adoption or quietly assume surveillance and headcount reduction are the real drivers. Confidence in handling ambiguity shows up in whether people experiment or wait for permission. Research tracking 48 million employee responses found that healthy workplace behaviours declined 11% year-on-year while unhealthy behaviours increased 13%, and this erosion is happening at exactly the moment organisations are asking people to take risks with unfamiliar tools. You cannot ask people to experiment openly in a culture that punishes uncertainty.
Organisations carrying trust deficits do not get a grace period when AI lands. They get acceleration. Surveillance concerns, which may have been latent in cultures where management relied on proximity-based oversight, crystallise fast when AI-enabled productivity monitoring enters the picture. Competency anxiety, the quiet fear of being exposed as less capable than a peer who adopted AI faster. Spreads through teams where psychological safety is already fragile. Decision-making opacity intensifies when people cannot tell whether a recommendation came from a manager, an algorithm. Or a combination of both. Microsoft’s 2026 Work Trend Index, drawing on 20,000 workers across 10 countries, found that organisational factors including culture, manager support, and talent practices account for twice the reported AI impact of individual effort alone. Culture is not a background condition in AI adoption. It is the primary variable.
Leaders who want to build cultures capable of absorbing AI at the speed the business requires need to normalise honest conversations that most organisations currently avoid. That means talking openly about what AI tools actually do well and where they produce unreliable outputs. It means establishing clear ownership of AI-informed decisions so that accountability does not dissolve into the algorithm. It means designing an approach to errors that treats mistakes as information rather than failure. Because teams that fear blame will hide AI-related errors rather than surface them.
The ethical dimension of AI culture does not live in policy documents. It lives in daily practice, in how leaders talk about the tools they use. Whether they acknowledge when their own work involves AI assistance, and how they handle the gap between colleagues who have embraced AI quickly and those who are struggling. Perceptyx research shows that 33% of employees report AI-related concerns. Whether those concerns become active disengagement depends almost entirely on leadership behaviour. Teams read these signals carefully, and what leaders do with AI tells people far more about the culture than any values statement pinned to the wall.
Hybrid Work and the Proximity Illusion
Many organisations believed they had strong cultures right up until the moment their people stopped sharing a building. When the pandemic forced distributed work at scale, those cultures did not travel. What remained was the shell of something that had always depended on proximity rather than principle. The ambient rituals of office life, the visibility of effort, the informal reinforcement of norms through physical co-presence. Hybrid work did not damage those cultures. It simply made visible what was already missing. Culture that lives in the building is not culture. It is atmosphere, and atmosphere evaporates the moment people stop breathing the same air.
By 2026, 74% of organisations operate in hybrid mode, which means this is no longer a transitional arrangement leaders can manage through patience and persuasion. Hybrid work is the operating condition, and culture must be engineered within it rather than despite it. The leaders who continue to defer deliberate culture-building, on the assumption that occasional office days will re-knit what distributed work has frayed, are misreading the situation entirely. Engineering culture in a hybrid environment requires explicit communication norms, clarity about what good performance looks like, and structured opportunities for connection that do not rely on physical coincidence to occur.
The surveillance response deserves particular scrutiny, because it is more widespread than most leadership conversations acknowledge. When leaders respond to hybrid work by monitoring keystrokes, mandating always-on cameras, or tracking login timestamps, they are not managing culture. They are making a statement about it. That statement reads as distrust, and employees receive it with considerable accuracy. Psychological safety, which is one of the strongest predictors of team performance, cannot coexist with the message that presence must be performed and monitored. Leaders who reach for surveillance tools in response to the discomfort of not seeing their people are treating visibility as a proxy for value. And they are embedding that assumption into the culture with every log they keep.
Outcome-based management offers the structural alternative, and its cultural significance extends well beyond its productivity implications. When a leader articulates clearly what success looks like, agrees those expectations with their team, and evaluates performance against outcomes rather than hours visible on a dashboard, they are making a cultural declaration. That declaration says the organisation trusts its people to deliver. It positions adults as professionals rather than subjects of oversight. The shift requires leaders to do harder work. Set precise expectations, hold honest conversations when results fall short, and resist the easier comfort of knowing that everyone is logged in. That precision is not a burden; it is the actual job of leadership.
Proximity bias compounds the challenge of inclusive culture in hybrid team environments, because its effects accumulate quietly in recognition, feedback, and promotion decisions long before leaders notice the pattern. Remote employees face a structural disadvantage in visibility. Not because they produce inferior work, but because leaders cannot notice their contributions through casual observation. Closing that gap is a deliberate leadership act. It requires meeting structures where remote participants hold equal standing. Recognition practices that surface distributed contributions proactively. And promotion criteria that measure documented output rather than remembered presence. Leaders must deliberately design an inclusive hybrid culture rather than expect it to emerge by default. They cannot delegate that responsibility to goodwill or good intentions.
The Shift from Title to Competence, and Why Legacy Hierarchies Push Back
The growing move toward skills-based hiring and development is not simply an HR methodology being trialled by progressive organisations. It is a direct cultural challenge to positional authority as the primary source of organisational influence and decision-making power. Deloitte has positioned the skills-based organisation as a new operating model for work and the workforce, signalling that this is a systemic redesign rather than an incremental policy update. When Deloitte’s Senior Partner and Lead in Workforce Transformation explicitly states that moving to skills ‘is a cultural change’ and requires organisations to redefine how they structure work, the implication becomes clear: this shift does more than upgrade talent management. It disrupts, at the level of organisational values, how organisations confer and legitimise authority.
The fundamental incompatibility between positional and competence-based authority is easier to describe than it is to resolve. In traditional hierarchical cultures, a title grants its holder authority regardless of demonstrated performance; decisions flow downward through seniority, and influence accumulates through tenure and rank. In a skills-based culture, influence is earned through what you can demonstrably do and how reliably you deliver it. As OrgShakers notes, traditional workforce models “rely on degree requirements, fixed hierarchies, and narrowly defined responsibilities,” and a skills-based model challenges each of those structural assumptions simultaneously. The two models cannot fully coexist without one subordinating the other; organisations that attempt to layer skills language onto an unchanged power architecture will find that the language shifts while the culture does not.
The resistance this generates at the middle management tier is not obstruction for its own sake. Leaders whose authority derives primarily from positional hierarchy are responding rationally to a genuine threat to the source of that authority. When competence becomes the primary currency of influence, a manager who has built their professional identity around rank and seniority faces something more significant than a new HR process. They face a direct challenge to their social standing, decision-making legitimacy, and self-concept within the organisation.
Senior leaders who dismiss this resistance as irrational are making a costly diagnostic error. As LACE Partners argues, people must adopt skills as a mindset rather than simply comply with a directive, and senior leaders cannot mandate that shift from above. Middle management represents the point where culture change most often stalls because skills-based models directly challenge positional authority at this level, while senior leaders have yet to demonstrate their commitment to the new approach through consistent action.
This points to one of the most underexamined failure modes in organisational transformation: the gap between declaring a skills-based culture and structuring power to match that declaration. Organisations routinely announce the shift, update their job frameworks, and invest in skills mapping tools, then discover that actual decisions still flow through the old authority channels. The corrective is not more communication; it is behavioural modelling from the top. A senior leader who defers to a more competent junior colleague in a domain where the junior holds genuine expertise is demonstrating the new culture. A leader who removes themselves from decisions where they lack relevant knowledge is demonstrating it. These behaviours are uncomfortable precisely because they require leaders to voluntarily relinquish the authority that positional hierarchy would otherwise protect.
The DEI dimension of this shift deserves direct treatment rather than a footnote. Removing title-based gatekeeping from development and promotion pathways is one of the most structurally durable mechanisms for building inclusive organisations, because it embeds fairness in process design rather than declaring it in communications. Skills-based hiring eliminates unnecessary barriers, including degree requirements and specific industry credentials that have historically excluded capable individuals from underrepresented groups. This matters particularly in the current environment, where DEI programmes framed as standalone initiatives face significant headwinds. A skills-based culture does not need a DEI label to produce equitable outcomes. Instead, organisations create greater fairness by designing systems in which people earn influence through their skills and gain access to opportunities through transparent criteria, making fairness a product of operational design rather than stated intent.
Ethics as Daily Practice, Not Policy Document
Every organisation has an ethics policy. Most of them are well-written, professionally formatted, and consulted primarily when something has gone wrong. New employees encounter them during onboarding, sign to confirm they have read them, and then watch carefully to see whether any of it resembles how decisions actually get made. Within a few months, sometimes weeks, the gap between the documented values and the lived reality becomes apparent. When that gap exists, people do not conclude that the policy is aspirational; they conclude that the policy is a legal defence mechanism, and they calibrate their behaviour accordingly.
The Culbertson analysis of ethical AI adoption identifies three structural layers of organisational culture: formal artefacts such as codes and values statements, normative behaviours embedded in everyday decision patterns, and reinforcement mechanisms including training and incentives. Ethical culture only becomes real, the research argues, when all three layers operate together. Most organisations have the first layer. Fewer have integrated the second and third. The result is ethics as documentation rather than ethics as practice, and the distinction is visible to everyone inside the organisation even if it is invisible in the annual report.
The drivers pushing ethics from optional to essential are structural rather than aspirational. Concerns about data privacy, AI governance, and information transparency are no longer edge cases that compliance teams manage quietly. McKinsey’s January 2025 workforce research found that only 1 per cent of company leaders describe their organisations as mature in AI deployment, while 92 per cent plan to increase AI investment over the next three years. That gap between stated commitment and actual readiness creates exactly the conditions in which ethical inconsistency becomes visible at every level. People notice when leaders conceal how they reached AI-assisted decisions, handle data differently when nobody is watching, or explain outcomes without explaining the reasoning behind them. They draw conclusions. Those conclusions shape the culture more powerfully than any policy document.
Embedding ethics into daily practice requires specificity about what that practice looks like. It means giving feedback honestly rather than diplomatically, because consistent honesty under social pressure is itself an ethical act. Handling data with the same rigour whether or not anyone is watching, because integrity that requires observation is not integrity. It means making AI-assisted decisions visible rather than obscuring the process behind a confident-sounding conclusion. It also means explaining the reasoning behind decisions rather than simply announcing outcomes, because transparent decision-making allows people to challenge decisions, learn from them, or trust them.
The connection between this kind of ethical consistency and psychological safety is direct and well-supported. Teams that observe their leaders making ethically consistent decisions under genuine pressure, not just comfortable ones, develop the confidence to raise concerns, challenge assumptions, and surface problems early. Teams that observe ethical inconsistency learn a different lesson: stay quiet, comply, and protect yourself. The first produces innovation and resilience. The second produces risk concealed beneath surface-level compliance.
Continuous feedback cultures sharpen this dynamic considerably. When feedback is frequent and embedded in everyday interaction rather than deferred to a formal annual cycle, ethical behaviour cannot be quarantined to review season. Every conversation becomes a site where honesty, fairness, and transparency are either practised or avoided. That frequency creates both the opportunity and the pressure to make ethics visible in the ordinary cadence of work, which is precisely where organisational culture is actually formed.
Measuring the Culture That Actually Matters
Most organisations believe they are measuring their culture. What they are actually measuring is sentiment. Engagement surveys capture how people feel about their job on the day they complete the form, and that is useful data, but it tells leaders almost nothing about whether the culture is producing the decision-making patterns, risk tolerance, and collaborative norms the organisation actually needs. A team can score highly on belonging and job satisfaction while simultaneously burying difficult information, avoiding accountability, and rewarding the wrong behaviours. The survey would not show any of that, because it was never designed to.
The move toward wellbeing indicators and pulse surveys is a genuine step forward. Continuous listening creates a richer picture than the annual check-in, and tracking shifts in sentiment over time at least alerts leaders to deterioration before it becomes a retention problem. The risk, however, is that leaders use this data to manage perception rather than diagnose behaviour. When the response to a low score is a communications campaign rather than a structural change, the measurement process itself becomes part of the cultural problem. Employees notice when their feedback generates a town hall and a revised values poster instead of anything that changes how decisions actually get made. Research confirms this directly: most organisations fail to translate survey findings into meaningful change, and when people experience that pattern repeatedly, they stop giving honest responses even when anonymity is guaranteed.
Practical culture diagnosis requires looking at what happens at the actual decision points that reveal whether stated values are real. Who gets heard in senior meetings, and who gets routinely interrupted or ignored? Whose ideas receive credit, and whose contributions get absorbed into someone else’s narrative? Which behaviours get excused or overlooked when the person exhibiting them is a high performer the business cannot afford to lose? How does the organisation treat mistakes when they become visible: as information to learn from, or as evidence to be managed away from reputations? None of these questions appear on a standard engagement survey. All of them tell you more about the real organisational culture than any sentiment score ever could.
The shift from annual appraisals to continuous feedback creates the conditions for better cultural measurement, but only when the feedback is honest, specific, and demonstrably acted upon. A feedback system that generates data without generating change does not improve culture; it demonstrates that the culture does not respond to input, which is itself a significant cultural signal. The administrative overhead of continuous feedback without follow-through actively erodes trust in the process.
Leaders should treat cultural measurement as diagnostic rather than performative. A high engagement score is not the goal; it is potentially useful evidence. The goal is to understand what the score is actually telling you about the gap between the culture you are declaring and the one your people are living inside every working day. That gap is where culture change either begins or gets permanently deferred.
What Leaders Must Do Differently: Behaviour-Level Commitments
Culture change does not begin in the boardroom strategy session where values get defined. It begins in the meeting room on a Tuesday afternoon when a mid-level manager dismisses a dissenting voice with a raised eyebrow. And no one in the room says anything about it. Gallup’s research shows that managers account for at least 70% of the variance in employee engagement scores. Which means the cumulative weight of those Tuesday afternoon moments, multiplied across every team in every department, is what your culture actually is.
The moments that feel insignificant are the moments that matter most, precisely because they happen without announcement and without an audience that expects performance. Whether a leader credits the person who originated an idea, how they respond when someone they respect makes a costly mistake, what they choose when values and convenience pull in opposite directions. These are the decisions that define culture far more reliably than any values statement ever will.
Four behaviour-level commitments separate leaders who genuinely shift culture from those who merely describe the shift they intend to make. The first is making decisions that are visibly consistent with stated values, even when that consistency carries a cost. The second is responding to challenge with curiosity rather than defensiveness, because leaders who treat pushback as a threat systematically destroy the psychological safety that honest conversation requires. The third is modelling the competence and ethical standards they require from others, rather than exempting themselves from those standards by virtue of seniority. The fourth is treating culture measurement as genuine feedback rather than a reputation management exercise. Organisations routinely conduct engagement surveys and then manage the communication around the results rather than acting on the substance. When employees observe that pattern consistently, they stop engaging honestly with measurement tools altogether, and leaders lose the only reliable signal they had.
Leaders operating in 2026 need to extend these commitments into two dimensions that most culture frameworks have not yet caught up with. The first is AI transparency. When leaders use AI tools to inform decisions, naming that influence openly is not a technical disclosure requirement; it is a cultural signal about honesty. The second is hybrid equity. Proximity bias, the structural advantage held by those physically present in an office, reproduces itself through informal communication, spontaneous decision-making, and visibility in ways that hybrid communication norms need to actively counteract.
Empathy and emotional intelligence deserve particular attention here because they are still widely treated as personality traits rather than competencies. They are not fixed. Leaders develop them through structured 360-degree feedback cycles, deliberate role-play, and coached reflection on real interactions. Organisations that measure these skills with the same rigour they apply to financial performance will build cultures that absorb pressure rather than fracture under it. The organisations that lead in the years ahead will not be the ones with the most sophisticated values statements. They will be the ones whose leaders close the gap between what they say and what they do, consistently, under pressure, at every level of the hierarchy, in the moments no one is watching.
The Case for Treating Culture as a Leadership Responsibility
Organisational culture is not a soft concept. It is the mechanism through which every strategy either gets executed or quietly resisted. The pressures bearing down on that mechanism in 2026 are more simultaneous and more consequential than at any recent point. AI adoption, hybrid work, skills-based disruption, generational complexity, and the ethics imperative are not arriving one at a time. They are arriving together, and the culture you have built, or failed to build, will determine how your organisation responds to all of them at once.
The leaders who will navigate this successfully are those who understand that culture changes at the behaviour level, one decision at a time, not at the announcement level. Every section of this analysis has pointed toward the same conclusion: what your culture actually is can be read in what behaviour gets rewarded, what dissent gets tolerated, and what norms hold when no one is watching.
Four commitments follow directly from that understanding. Audit the gap between your espoused culture and your enacted culture by examining what behaviour actually gets rewarded in practice. Extend your cultural framework explicitly to cover AI governance and hybrid communication norms, because most existing frameworks do not. Replace sentiment measurement with behaviour observation at key decision points. And make one visible, behaviour-level commitment this week that demonstrates the culture you are trying to build. Because your people are not reading your values statement. They are reading you.
Leave a Reply