Does Culture Really Start at the Top? What 10 Years of Research Says

Does culture start at the top? The short answer

Yes, culture starts at the top, but it does not stay there on its own. A decade of research shows that senior leaders' behaviour is one of the strongest forces shaping organisational culture, for better and for worse. Yet leaders change culture less by being copied than by deciding what gets rewarded, tolerated and paid attention to. Their signal then travels through middle managers and peer norms. Culture change works best when it is top-first and top-through.

Few phrases are repeated in boardrooms as often as "culture starts at the top." It is said by CEOs launching transformation programmes, by consultants selling leadership offsites, and by employees explaining why nothing ever changes. Lately, a counter-movement has started calling it a myth. Culture, they argue, lives in teams, in middle management, in everyday interactions, not in the executive suite.

Both camps are partly right. And the difference matters, because where you start a culture change decides how much time, money and credibility you burn before anything moves.

At PeopleKult, we work on culture as a behavioural design problem. So instead of adding another opinion, we went back to the research of the last ten years: studies on CEOs and culture, on how behaviour "trickles down" hierarchies, on engagement cascades, and on how social norms form and lock in. This article summarises what that evidence says, where the popular version of "start at the top" goes wrong, and what a more precise model looks like for leadership teams who want culture change to stick.

What the research says about senior leaders and culture

The link between the top of an organisation and its culture is one of the better-documented findings in organisational science. Five streams of research point in the same direction.

1. CEOs leave a measurable imprint on culture

In a study of 32 technology companies, Charles O'Reilly, Jennifer Chatman and colleagues found that CEO personality shapes a firm's culture, and that culture in turn predicts revenue growth, market valuation, reputation and employee attitudes. A later study by O'Reilly, Xubo Cao and Donald Sull went much bigger. Analysing 1.2 million Glassdoor reviews, they found a significant correlation between CEO personality and how employees describe their company's culture.

Executives see it the same way. In a large survey of North American executives by John Graham, Campbell Harvey and colleagues, 52% said culture is primarily set by the current CEO, and boards mainly influence culture by choosing who that CEO is. In the published version of the study, 92% of 1,348 executives believed improving their culture would increase firm value, and inattentive leaders were among the main obstacles they named.

2. Behaviour trickles down the hierarchy

Researchers call it the trickle-down effect: what happens at one level of the hierarchy shows up at the next level below. A review of 62 empirical studies found that leaders' traits, emotions, attitudes and behaviours tend to produce similar responses in their followers and teams. A 2018 study in the Journal of Applied Psychology added direct evidence that social learning drives this process: when supervisors trusted their own managers, they behaved in more trusting ways with their teams, and performance rose.

The effect runs in both directions, good and bad. One study in the Academy of Management Journal found that mistreatment travels down through displaced aggression: managers who feel disrespected by those above them pass the disrespect on to those below.

3. Engagement cascades through each layer

Gallup's data shows the same pattern for engagement. Managers who report to highly engaged leadership teams are 39% more likely to be engaged than those reporting to disengaged leadership teams. Employees who report to highly engaged managers are 59% more likely to be engaged. Executives influence managers, and managers influence the front line.

4. Humility at the top creates empowerment below

A study of 63 companies by Amy Ou and colleagues traced how CEO humility cascades through the organisation. Humble CEOs led in more empowering ways, which made their top teams more integrated. Middle managers in those companies perceived a more empowering climate and reported higher engagement, commitment and performance.

5. Toxic cultures usually have toxic leadership behind them

Donald and Charles Sull analysed more than 1,000 studies on toxic workplaces. Their conclusion: leadership was consistently the best predictor of a toxic culture, ahead of social norms and work design. Leaders cannot fix a culture without holding themselves and their peers accountable for toxic behaviour.

Taken together, the verdict is clear. If you are a senior leader, your behaviour is not one input among many. It is one of the loudest signals in the system.

Why employees copy the people at the top

Knowing that behaviour flows downward is useful. Knowing why it flows is what lets you design for it. Behavioural science points to three mechanisms.

Prestige: we learn from people who look successful

Humans are selective social learners. Cultural evolution researchers Joseph Henrich and Francisco Gil-White proposed that we preferentially copy people who are admired, deferred to and paid attention to by others, a strategy known as prestige-biased learning. Later experiments, including a 2020 study in Scientific Reports, found that people lean on prestige most when they cannot see directly what leads to success.

That describes most organisations perfectly. Employees rarely know the real rules for getting ahead. So they watch the people who obviously got ahead, and copy them. One nuance matters here: some research suggests prestige is partly domain-specific. A CEO admired for winning deals will be copied on how to win deals, not automatically on how to treat people.

Uncertainty: the boss is the best available signal

Organisations are full of ambiguity. Is it safe to disagree in this meeting? Do deadlines really matter more than quality? Does "work-life balance" mean anything here? When the rules are unclear, the behaviour of those with power over our careers becomes the most reliable source of information. Social information processing theory describes this: people build their sense of "how things work here" from the cues around them, and cues from the top carry extra weight.

Survival: hierarchy rewards alignment

In any hierarchy, the people above us control resources we need: promotions, projects, budgets, protection. Aligning with their behaviour is a rational strategy for staying safe and moving up. This is why the tolerance signal is so powerful. When a senior leader ignores a high performer's bullying, every observer updates their model of what is actually rewarded.

Words versus deeds

All three mechanisms explain one robust finding: people copy what leaders do, not what leaders say. A meta-analysis on behavioural integrity, the alignment between a leader's words and actions, found it has strong effects on trust, commitment and performance. Part of the reason is clarity. When words and deeds match, employees know exactly what is expected. When they diverge, employees follow the deeds and stop believing the words.

Why "start at the top" is not enough

If the evidence for leader influence is so strong, why do so many top-down culture programmes fail? Because the popular version of the idea assumes something the research does not support: that employees simply imitate the CEO, and culture follows. Three findings complicate that picture.

Most employees experience the top through their manager

Gallup estimates that managers account for at least 70% of the variance in team engagement. For most employees in a large company, the CEO is a face on a town hall screen. The manager is the person who decides whether their idea gets heard on Tuesday morning. Whatever signal leaves the executive floor is amplified, distorted or blocked by that layer.

The Sulls' toxic culture research makes the same point from the other side: middle managers can create toxic pockets even under a healthy top team, and some norms take on a life of their own, independent of the leader who started them.

Leaders shape culture more through systems than through imitation

One of the most rigorous studies on this question followed 2,572 soldiers across three levels of hierarchy during a combat deployment. The researchers found only limited support for simple trickle-down, where people just copy the leader above. They found much broader support for a different mechanism: senior leaders embed shared expectations in the culture of the units below them, and they shape how their subordinate leaders behave. That culture, not direct imitation, is what reaches the front line.

This matches what Edgar Schein described decades ago. Leaders embed culture through what they pay attention to, measure and control, how they react to crises, how they allocate resources, and whom they hire, promote and fire. These signals reach people who have never met the CEO.

The bigger the organisation, the weaker the direct effect

Research on the "CEO effect" shows that CEOs have become more influential on firm performance over recent decades. But a large replication in Strategic Management Journal found that the larger the firm and the longer the CEO's tenure, the smaller the share of performance the CEO explains. In a 50-person scale-up, the founder's behaviour is the culture. In a 20,000-person group, it is one powerful signal competing with thousands of local ones.

Change can also come from below

Finally, culture is not only transmitted from the top. Damon Centola's experiments found that a committed minority can overturn an established social convention once it reaches a critical mass, around 25% of the group in their setting. That exact number should not be applied to every organisation, since real workplace norms are more complex. But the principle holds: peers are a second engine of culture change, and they can work with the top or against it.

The timing effect: leaders matter most at the beginning

The most practical finding in this body of research is about timing, and it rarely appears in culture programmes.

Behavioural economists Simon Gächter and Elke Renner ran an experiment in which groups decided, round after round, how much to contribute to a shared project. In some groups, a leader contributed first and everyone saw it. Leaders strongly shaped what followers believed and how much they contributed at the start. But as the rounds went on, followers paid more attention to what their peers had done than to the leader's current behaviour. Once a pattern had formed among peers, the leader could hardly correct it.

Translate that into organisational life and you get a clear rule: leader behaviour has its greatest leverage when beliefs are still forming. After that, peer norms take over and become self-reinforcing.

Those formative windows are predictable:

• A new CEO or leadership team's first months, when everyone is watching to see what is really valued.

• Mergers and restructurings, when the old rules no longer apply and the new ones are not yet clear.

• Crises, when people learn more from one leadership decision than from a year of values workshops.

• Onboarding, when every new hire forms their first model of "how things work here."

• The launch of a culture initiative, when early behaviour from the top decides whether employees treat it as real or as another poster campaign.

The implication is uncomfortable but useful. A leadership team that behaves inconsistently in the first weeks of a change effort is not just losing time. It is helping peers lock in the old pattern, which then becomes much harder to shift.

A better model: top-first, top-through

So does culture start at the top? The evidence supports a more precise statement:

Senior leader behaviour is the highest-leverage starting point and a necessary condition for culture change, but not a sufficient one. Leaders change culture through three channels: what they model, what they embed in the system, and how their managers translate it. Culture change works when it is top-first and top-through.

Channel 1: Modelling (what leaders visibly do)

This is the channel most people think of. How leaders run meetings, react to bad news, treat people with less power, spend their time and admit mistakes. It works through prestige and uncertainty, and it is strongest in formative moments. Its limit is visibility: in a large organisation, most employees only see a fraction of it.

Channel 2: Embedding (what leaders reward, tolerate and pay attention to)

This is the channel the research suggests is underrated. Every promotion, every budget decision, every metric on the leadership dashboard and every behaviour a leader walks past sends a signal that travels far beyond the people in the room. Embedding is how a CEO's values reach a team in another country who have never met them. It is also where most misalignment hides. A leadership team can model collaboration in town halls and still promote the people who hoard information.

Channel 3: Translating (how managers carry the signal)

Middle managers decide what the signal from the top means in daily work. If they are engaged, equipped and held to the same standard, they amplify it. If they are exhausted, sceptical or rewarded for something else, the signal dies in the middle. Gallup's cascade data and the Sulls' research on toxic subcultures both point here.

The fourth force: peers

Beneath all three channels sit peer norms. They are the reason a culture feels stable, and the reason early leader behaviour matters so much. Top-first, top-through means working with peers deliberately: identifying the people others already look to, and helping a committed group reach critical mass so the new behaviour becomes the normal one.

"Top-first" answers the question of where to start. "Top-through" answers the question that most culture programmes skip: how the change actually reaches the people who do the work.

What this means for your leadership team

Here is how we translate the research into practice when we work with leadership teams. None of these steps requires a large budget. All of them require the top team to go first.

1. Define behaviours, not values

"Respect" and "accountability" cannot be copied, because nobody can see them. Behaviours can. Translate each value into two or three observable leadership behaviours: "I ask for dissenting views before I give mine", "I name my own mistakes in team meetings", "I address disrespect within 24 hours." If a new employee could not spot the behaviour in a meeting, it is not specific enough.

2. Audit what you reward and tolerate

Look at the last ten promotions, the last three exits of high performers, and the behaviours your leadership team has walked past this quarter. Ask a simple question: if employees only had this data, what would they conclude our culture really values? The gap between that answer and your values statement is your real culture change agenda.

3. Make leader behaviour visible on purpose

Because modelling is limited by visibility, good behaviour that nobody sees changes nothing. Leaders can narrate their decisions ("I'm changing my view because of what Sara raised"), share their own development goals, and show up where the culture is actually lived, not only on stage.

4. Use the formative windows

Map the moments in the next six months where beliefs will form: a new leader joining, a reorganisation, a difficult decision, a cohort of new hires. Plan the leadership behaviours for those moments in advance. One visible decision in a crisis does more than a year of internal campaigns.

5. Equip and align the middle

Give managers the same behavioural standards as the top team, the skills to apply them, and the same accountability. Then check whether their incentives point in the same direction. A manager rewarded only for short-term targets will translate any culture message into "hit your numbers."

6. Build a peer coalition

Identify the informal influencers in each part of the organisation, the people others already go to for advice. Involve them early and give them a real role. Your aim is a committed group large enough that the new behaviour starts to feel normal.

7. Measure the cascade, not just the average

Company-wide engagement scores hide where the signal breaks down. Look at culture and engagement data level by level and team by team. If the top team's scores are strong and frontline scores are weak, the problem is not your values. It is the translation. Tools like AI-based people analytics can show where in the hierarchy the signal is getting lost.

If you want a low-cost starting point, we have written a step-by-step guide on how to start a culture change cheap and fast.

Frequently asked questions

Does organisational culture really start at the top?

Largely, yes. Research links CEO personality and senior leader behaviour to culture, engagement and even financial performance. But leaders shape culture mainly through what they reward, tolerate and embed in systems, and through their managers. Starting at the top is necessary, not sufficient.

How do leaders influence organisational culture?

Through three channels: modelling (what they visibly do), embedding (what they reward, measure, tolerate and pay attention to) and translating (how middle managers interpret and pass on their signal). Peer norms then reinforce or resist what comes from above.

What is the trickle-down effect in leadership?

It is the finding that attitudes and behaviours at one level of the hierarchy tend to reappear at the level below. Trust, fairness, engagement, and also mistreatment have all been shown to trickle down from senior leaders to managers to employees.

Can culture change from the bottom up?

Yes, peer groups can shift norms once a committed group reaches critical mass. But bottom-up change rarely lasts if senior leaders keep rewarding the old behaviour. The most durable change combines both directions.

What role do middle managers play in culture change?

A decisive one. Gallup estimates managers account for at least 70% of the variance in team engagement. They translate the signal from the top into daily work, so culture change that does not equip and align them usually stalls in the middle.

You will find more answers in our article on the 20 questions every company is asking about work culture.

Conclusion: start at the top, then follow the signal down

"Culture starts at the top" is neither a cliché nor a myth. It is an incomplete truth. The research of the last decade shows that senior leaders are among the most powerful forces in any culture, and that their behaviour, good or bad, flows down the hierarchy. It also shows that this influence travels mostly through systems and managers, that it is strongest in formative moments, and that peers decide whether a new behaviour becomes the norm.

For leadership teams, the lesson is simple to state and hard to practise. Go first, be consistent, and then make sure your signal survives the journey to the front line.

Want to know where your culture signal breaks down? PeopleKult's Culture Accelerator helps leadership teams define the behaviours that matter, align what they reward with what they say, and track how culture cascades through every level of the organisation. Book a conversation with our team.

References

• Centola, D., Becker, J., Brackbill, D. & Baronchelli, A. (2018). Experimental evidence for tipping points in social convention. Science, 360(6393), 1116-1119.

• De Cremer, D., van Dijke, M., Schminke, M., De Schutter, L. & Stouten, J. (2018). The trickle-down effects of perceived trustworthiness on subordinate performance. Journal of Applied Psychology, 103(12), 1335-1357.

• Gächter, S. & Renner, E. (2018). Leaders as role models and 'belief managers' in social dilemmas. Journal of Economic Behavior & Organization, 154, 321-334.

• Gallup. Managers account for 70% of variance in employee engagement.

• Gallup. The cascade effect: managers are engaged in their jobs.

• Graham, J. R., Grennan, J., Harvey, C. R. & Rajgopal, S. (2022). Corporate culture: Evidence from the field. Journal of Financial Economics, 146(2), 552-593.

• Henrich, J. & Gil-White, F. J. (2001). The evolution of prestige. Evolution and Human Behavior, 22(3), 165-196.

• Keller, T., Glaum, M., Bausch, A. & Bunz, T. The CEO in context technique revisited: a replication and extension of Hambrick and Quigley (2014). Strategic Management Journal.

• O'Reilly, C. A., Caldwell, D. F., Chatman, J. A. & Doerr, B. (2014). The promise and problems of organizational culture: CEO personality, culture, and firm performance. Group & Organization Management.

• O'Reilly, C. A., Cao, X. & Sull, D. Follow the leader: how a CEO's personality is reflected in their company's culture. Stanford GSB Insights.

• Ou, A. Y., Tsui, A. S., Kinicki, A. J., Waldman, D. A., Xiao, Z. & Song, L. J. (2014). Humble chief executive officers' connections to top management team integration and middle managers' responses. Administrative Science Quarterly, 59(1), 34-72.

• Quigley, T. J. & Hambrick, D. C. (2015). Has the "CEO effect" increased in recent decades? Strategic Management Journal, 36, 821-830.

• Schaubroeck, J. M. et al. (2012). Embedding ethical leadership within and across organization levels. Academy of Management Journal, 55(5), 1053-1078.

• Simons, T., Leroy, H., Collewaert, V. & Masschelein, S. (2015). How leader alignment of words and deeds affects followers: a meta-analysis of behavioral integrity research. Journal of Business Ethics, 132(4), 831-844.

• Sull, D. & Sull, C. How to fix a toxic culture. MIT Sloan Management Review.

• Wang, Z., Xu, H. & Du, C. (2015). The trickle-down effect in leadership research: a review and prospect. Advances in Psychological Science, 23(6), 1079-1094.

• Wo, D. X. H., Ambrose, M. L. & Schminke, M. (2015). What drives trickle-down effects? A test of multiple mediation processes. Academy of Management Journal, 58(6), 1848-1868.