A founder I know can name the week her company's culture changed. It was not the week they wrote the values down.
It was the week a manager she had never spoken to told a new starter, in her absence, that a customer complaint could wait until Monday. Nobody had written that policy. Nobody would have defended it out loud. It was simply what happened when she was not there.
When I interviewed Trip Hawkins, who founded Electronic Arts after an early career at Apple, he described watching the same problem at a scale most founders never see. "This conviction was put to the test when I joined Apple, a company that, at my arrival, was a tight-knit group of 25 office workers. Within four years, our ranks swelled to 4,000, presenting a first-hand view of the challenges and changes that accompany rapid organizational growth."
Culture is the subject founders discuss most confidently and understand least well. It is treated as something you design, announce and then maintain, like a brand. It is nothing of the sort. It is the residue of behaviour, it is set largely by what senior people actually do, and it degrades at sizes that are entirely predictable.
01What is culture actually made of?
Behaviour, repeated, observed and copied. Nothing else.
Marc Randolph, who co-founded Netflix and was its first chief executive, put this to me about as bluntly as it can be put. "Culture is not aspirational, it's observational. It's not something you and your co-founders sit down, dream-up, put into PowerPoint... Culture, simply, is how you behave and how you treat your co-founders, employees, and customers."
The value of that formulation is that it makes culture measurable. If culture is a set of aspirations, you can only survey people about how they feel. If culture is observable behaviour, you can look at what happened: who got promoted, which client was fired, whose bad quarter was forgiven and whose was not.
Gilles Carbonnier, vice-president of the International Committee of the Red Cross, gave me the standard against which most corporate culture statements fail. "We walk the talk: our only agenda is humanitarian, and we perform those humanitarian tasks in a neutral, independent and impartial manner," he told me, describing an organisation whose credibility depends on its principles and its behaviour being the same thing, checked daily where the gap between the two can get people killed. Most companies do their cultural work entirely on the stated side of that gap, which is why so much of it changes nothing.
02Why does the founder's behaviour travel so far?
Because status governs imitation, and in a young company the founder holds nearly all of it.
Will Storr, who wrote The Status Game, explained the mechanism to me in terms that had nothing to do with business. "the criteria for identifying these high-status individuals involve several cues: similarity, skill, success, and prestige... we mimic these individuals' behaviours, preferences, and even speech patterns."
This is why the perks conversation is a distraction. As Melissa Daimler, who has led people and learning functions at Adobe, Twitter, WeWork and Udemy, told me: "I don't think culture is the bean bag or the Ping-Pong tables or the free food or the happy hours... I do not think those things define what a culture really is." The bean bag communicates nothing because nobody's status depends on it. Whether the founder answers a difficult email on the day it arrives communicates a great deal, because forty people are watching to learn what is normal here.
It also means the founder's worst habits scale faster than their best intentions. Cutting a corner under deadline pressure once, visibly, does more cultural work than a quarter of internal communication about quality, in the same way that the hiring bar itself is a statement about what a business will tolerate.
03Where does founder-led culture break as the company grows?
At the point where the founder can no longer be personally present in the decisions that transmit the culture.
Tony Hsieh, who built Zappos, described the moment precisely. He had hired on instinct in the early days, using the test of whether he would choose to spend time with a person socially. "As a small business, this was fine, but as we grew? we ran into scaling problems... That led us to create our 10 core values, instilled into our culture."
That is the first break, and it usually arrives somewhere between thirty and fifty people: the founder is no longer in every hiring decision, so behaviour begins to propagate through people who learned it second-hand, which is really the same design problem that delegation always turns out to be. The second break comes later, when informal reputation stops working because nobody knows everybody, and the shared memory of how things are done here has no natural carrier.
A culture that requires the founder's presence is not a culture. It is a personality, and personalities do not scale.
Ron Shaich, who built Panera Bread to over a hundred thousand employees, was asked repeatedly to appear on Undercover Boss and always refused, for a reason that is really a statement about system design. "value itself is a byproduct," he told me. "So what should you focus on? The job of an entrepreneur or a competitive leader is to create the conditions that make your organization the better competitive alternative." Culture, at that scale, is one of those conditions, not an output of the founder's presence, and has to be built as deliberately as any other advantage.
04How do you codify culture without killing it?
By describing behaviour precisely enough that a breach is visible, and by codifying judgement rather than procedure.
Harvey Whitehouse, Professor of Social Anthropology at Oxford, has spent his career on how small groups transmit identity to large populations, and the mechanism he described to me is exactly what a scaling company needs. "Routinizing rituals... makes deviations from the standard script easy to detect. And this means that when people step out of line, they can be sanctioned... The power of routinization is that it allows you to spread a shared set of identity markers to a large population, and to do so very quickly."
The operative phrase is easy to detect. A value that cannot be breached is not a value. "Integrity" fails this test, because no organisation claims the opposite. "We tell the client about a problem before they find it" passes, because you can point at an occasion when somebody did not.
Which is also why honesty in the drafting matters more than ambition. Adeo Ressi, who runs the Founder Institute and has watched thousands of startups pass through it, put the standard bluntly: "You can't be successful unless you're genuine as a human being and honest... The startups in our world that are really successful are both genuine and honest." A values statement written by people who are neither describes an organisation nobody recognises, which fails more expensively than writing nothing.
05Where the written culture and the real one split
Four things, and the first three are self-inflicted.
The statement describes the company you wish you were. Written values that describe an aspiration rather than a practice teach employees that official language is not connected to reality. This is worse than having nothing written, because it establishes that the gap between words and behaviour is acceptable here.
Codification becomes accumulation. Every incident produces a new rule, nobody ever removes one, and within three years the handbook describes an organisation that does not trust its own staff. The cure is making deletion as routine as addition, annually, by name.
Senior hires import their previous culture wholesale. A director arriving from a large company brings a working model of normal that was formed elsewhere and is invisible to them. Within two quarters their team behaves like their last employer. This is entirely manageable, but only if somebody has said out loud what the behaviours are.
Nobody is ever removed for behaviour alone. The single fastest way to destroy a stated culture is to retain a high performer who breaches it. Everyone watches that decision, and everyone draws the correct conclusion.
06Is any of this actually necessary?
The honest objection is that good cultures mostly happen without being managed, and that formalising them is the beginning of the decline.
There is real evidence for it. When I interviewed Natasha "Mo" Hunt, the England rugby international, she described her club side in terms that would horrify most culture consultants. "We don't actively discuss culture because we're surrounded by good people. When you have good people, the culture naturally thrives on its own."
She is right, and the qualifying condition is in her sentence. A squad is small, its members are selected with extraordinary care, and everyone can see everyone else work. Under those conditions culture genuinely does propagate on its own, and codification is a solution to a problem you do not have. Plenty of thirty-person businesses are in exactly this position and should leave it alone.
The condition fails when you can no longer see everybody, or when you are hiring faster than the existing group can absorb newcomers. At that point you are choosing between an articulated culture and an accidental one, not between an articulated culture and the good thing you had.
07Every promotion is a values statement
What would you find if you looked at your last ten decisions made without you in the room? Write down what happened, then compare it against what you would say your culture is. The gap is your actual culture, and it is the only version your employees have ever experienced.
Then pick three behaviours from the good half of that list, describe them in a sentence each that somebody could be shown to have breached, and use them the next time you promote someone. Culture at scale is not a document or a set of feelings. It is the accumulated evidence of which behaviours were rewarded, and every promotion is a louder statement of values than anything you will ever write down.