Walk into a twenty-year-old company and you can feel its culture before anyone explains it, in how meetings start, who interrupts whom, whether bad news travels up. None of that was decreed. It got learned, repeated until it went invisible, then handed to every new joiner as "how things are done here." Knowing how that happens is what separates wishing your culture were different from being able to move it.

The quick version

  • Culture forms by learning. A group hits a problem, tries a response, it works, and the response quietly hardens into a shared assumption nobody argues with anymore. Edgar Schein called these assumptions the real culture; the visible stuff (offices, slogans) sits on top.
  • Founders and early leaders shape it most. What the first leaders pay attention to, reward, and panic about teaches the group what matters, far more than any values statement.
  • It persists through people. Organisations attract, hire and keep people who fit, and the ones who don't fit leave. Over time that makes the place more like itself, which is stabilising, and also a diversity risk.
  • So you can't change culture by announcing new values. You change it by changing what leaders consistently notice, measure and reward, the same levers that built the old culture in the first place.

The idea in depth: culture is learned, then forgotten

The foundational map here belongs to Edgar Schein, the MIT Sloan psychologist who, more than anyone, made culture something you could study rather than just sense. In Organizational Culture and Leadership he defined culture in three levels. At the surface are artifacts, what you can see and hear: the office layout, the dress code, the rituals, the public handbook. Below that sit espoused values, what the organisation says it believes, the words on the wall. And underneath, doing the actual governing, are basic underlying assumptions, beliefs so taken-for-granted that members no longer notice holding them (MIT Sloan summary).

The crucial mechanism is how an assumption gets to the bottom layer. Schein's definition is that culture is the pattern of assumptions a group "invented, discovered, or developed in learning to cope with its problems of external adaptation and internal integration", a pattern that "has worked well enough to be considered valid" and is therefore "taught to new members as the correct way to perceive, think, and feel" about those problems. The phrase that matters is worked well enough. A solution that succeeds gets repeated; a repeated solution stops being a choice and becomes simply reality.

flowchart TD
  A(["A group hits a problem"]) --> B(["Someone proposes a response
(at first, just an opinion)"])
  B --> C{"Did it work?"}
  C -->|"Yes, repeatedly"| D(["It becomes an espoused value"])
  D --> E(["Stops being questioned,
now a basic assumption"])
  E --> F(["Taught to newcomers as
"how things are done here""])
  C -->|"No"| G(["Dropped, try again"])
  G --> B
How a one-off response sinks from opinion to invisible assumption, Schein's learning model of culture. Leaders Loop

So the move is: stop auditing culture by reading your values page and start reading your assumptions. Ignore the slogans and watch the cracks, what gets someone praised in a hallway, what makes a manager flinch, which rule everyone breaks and which one nobody dares to. A practical test: when a new joiner does something "wrong," what unwritten rule did they break? Write those down for a month; that list is closer to your real culture than anything HR has published. (This connects directly to defining & embedding values, the gap between espoused and lived values is the gap between the middle and bottom layers.)

Why founders and early leaders matter so much

If culture is learned, the obvious question is: learned from whom? Schein's answer, set out in his paper "The Role of the Founder in Creating Organizational Culture", is that the earliest leaders have wildly disproportionate influence. A founder doesn't just start the company; they bring the first set of beliefs about how to succeed and how to treat people, and because the group has no prior history, the founder's assumptions become the assumptions when their bets pay off.

The mechanism isn't grand speeches. Schein argued leaders embed culture through a handful of mundane, repeated behaviours, what he called primary embedding mechanisms: what leaders pay attention to and measure; how they react in a crisis; how they allocate resources, rewards and status; and who they recruit, promote and quietly push out. The group reads these signals far more closely than any mission statement, because actions under pressure reveal what is really valued.

A team learns its culture not from what leaders say they value, but from what leaders reward, panic about, and promote.

So the move is: treat your own attention as the strongest culture lever you have. If you say you value learning but only ever measure delivery, the team learns that delivery is what counts and "learning" is decoration. Pick the two or three behaviours you genuinely want to be cultural and ask, ruthlessly: do I measure them, reward them, and promote the people who model them? A single promotion teaches more about your culture than a year of all-hands. If the answer is no, your espoused value is sitting on the middle layer and will never sink to the bottom. (More on the leadership behaviours that do the embedding in leadership styles & models.)

An honest limitation. The founder-centric story is cleanest in young companies. In a large, old, multinational organisation, culture is no longer one person's shadow, it is a contested patchwork of national, occupational and divisional subcultures that can diverge sharply from the centre. Schein himself stressed this in his later work. So "shape it from the top" gets weaker the bigger and older you are; in a mature firm, the realistic unit of culture change is often the team or the function, not the whole enterprise.

Why culture persists, even after the founders leave

Here is the part leaders underestimate: culture doesn't need its creators to survive them. The clearest explanation comes from organisational psychologist Benjamin Schneider, whose 1987 paper "The People Make the Place" proposed the Attraction–Selection–Attrition (ASA) cycle. The argument is simple and a little unsettling: an organisation becomes defined by the kinds of people in it, and the people in it are filtered by three forces. People are attracted to organisations whose values look like their own; organisations select those who fit their norms; and those who don't fit eventually leave, attrition. Run that cycle for years and the place becomes steadily more homogeneous, more like itself, with or without the original leaders.

flowchart LR
  A(["Attraction
like-minded people apply"]) --> B(["Selection
you hire those who 'fit'"])
  B --> C(["Attrition
the misfits leave"])
  C --> A
  C -.-> D(["Result over time:
a more homogeneous,
self-reinforcing culture"])
Schneider's ASA cycle, the engine that keeps a culture stable long after its founders are gone. Leaders Loop

This is why culture is so durable and so hard to shift: it is held in place not by rules but by who is in the room. There is a real danger buried in that. The same process that gives an organisation a strong, stable identity is the one that erodes its diversity of thought, and a workforce that has selected itself into sameness adapts badly when the market changes. Schneider's insight cuts both ways: stability and stagnation come from the same engine.

So the move is: if you want a culture that holds but doesn't ossify, intervene at the "selection" step deliberately. That means hiring for culture add rather than culture fit, asking "what does this person bring that we're missing?" rather than "will this person feel like us?" It means watching who leaves and asking whether your attrition is quietly editing out dissent. And it means being honest that a "great culture fit" verdict is often just "reminds me of us." (This is the practical hinge into diversity, equity & inclusion and onboarding & ramp, where the ASA cycle is either reinforced or interrupted.)

A worked example

Take a software company, call it Harbor, founded by two engineers who shipped their first product by working through a weekend outage themselves, refusing to wake anyone else. (Illustrative example; the people and figures are invented to teach the model, not a real company.) That weekend felt heroic, it worked, and it quietly taught the group an assumption: real commitment means absorbing the pain yourself rather than escalating. Nobody wrote it down. It just became how things were done.

Five years on, Harbor has 200 people and a values page that says "we communicate openly." But engineers still sit on problems until they become fires, because the bottom-layer assumption, escalating looks weak, is the one actually running. Through the ASA lens, it gets worse: Harbor has been hiring people who admire that stoicism, and the few who pushed for earlier escalation found the place exhausting and left. The culture is now self-reinforcing.

A new VP of Engineering wants psychological safety and early escalation. The textbook-wrong move is to announce a new value and update the wiki, that just adds another espoused value on top of an unchanged assumption. The move that follows the model is to work Schein's embedding mechanisms. She changes what gets rewarded: the first engineer to raise a risk early gets visible, specific public credit, even when the risk turned out to be nothing. She changes how she reacts in a crisis: when the next outage hits, she opens by thanking whoever escalated, not by hunting for blame. And she changes selection: the next two hires are people who instinctively over-communicate, deliberately adding what the culture had filtered out. Within a couple of cycles, escalating early stops being a career risk and starts being the obvious thing to do. The assumption shifts, because the signals that built the old one were turned to build a new one.

Frequently asked questions

Can you actually change an organisation's culture, or are you stuck with it?

You can, but not by decree, and not fast. Because culture lives in assumptions formed by repeated experience, it changes only when leaders give the group new experiences that contradict the old assumption, and reward the new behaviour consistently enough that it, too, becomes "just how we do things." Announcing values changes the middle layer; changing what you measure, reward and promote is what eventually reaches the bottom layer. Expect it to take cycles, not quarters.

What's the difference between culture and values?

Values are what an organisation says it believes, Schein's "espoused values," the middle layer. Culture, in the sense that governs behaviour, is the layer below: the assumptions people act on without thinking. When the two match, the values are real; when they diverge, the gap itself is a cultural signal (it teaches people that the stated values are for show). Values are an input you can write; culture is an outcome you can only shape.

Why does culture survive when the founders and original team have all gone?

Because it is held in place by who gets hired and who stays, not just by who started it. Schneider's Attraction–Selection–Attrition cycle means an organisation keeps drawing in and retaining people who fit its existing norms, so the culture reproduces itself through successive generations of staff. The founders set the pattern; the hiring and attrition process maintains it automatically.

Is a "strong" culture always a good thing?

No. A strong, cohesive culture aligns people and speeds decisions, but the same forces that make it strong, selecting for fit, losing the misfits, reduce the diversity of thought that helps an organisation adapt when conditions change. A culture can be strong and wrong, and the stronger it is, the harder it resists the correction. Strength is a feature when the world is stable and a liability when it shifts.

Where should a leader start if the culture feels off?

Start by diagnosing the bottom layer, not rewriting the top one. Spend a few weeks noticing what actually gets rewarded, what people apologise for, and which unwritten rules newcomers trip over, that tells you the real assumptions. Then pick one or two you want to change and adjust the signals you personally control: what you measure, how you react under pressure, and who you promote. Culture change is led by example at the level of the embedding mechanisms, not the slogan.

Related in the Toolkit

Understanding how culture forms is the groundwork for almost everything in this area, from turning words into lived norms (defining & embedding values) to keeping the ASA cycle from quietly narrowing who belongs (diversity, equity & inclusion).

Where to go next