We've Had Enough Entrepreneurship. Let's Build a Bureaucracy.

I have yet to meet the entrepreneur whose ambition was to build a sizeable HR department. Nobody sits around the kitchen table in the early days of a company dreaming that, if things go particularly well, they might eventually employ enough people to need policies governing how those people employ other people. Compliance probably does not feature heavily in the dream either, and I doubt many founders have celebrated their first Procurement approval matrix with champagne.

Yet if the company succeeds, some version of all of this is probably coming. There will be more employees, bigger customers, larger contracts, more money, more sensitive information and considerably more opportunity for somebody to do something spectacularly stupid. The founders can no longer know everyone personally, decisions have consequences well beyond the person making them, and "I trust Sarah, she knows what she's doing" becomes increasingly difficult to defend as an enterprise governance model. At some point the company needs HR, although I remain unconvinced by the comforting corporate fiction that HR is there to protect the employees. It is there because employing people creates complexity, obligations and risk for the organisation, and once there are enough people, somebody has to manage all three.

Compliance, Legal, Finance, Procurement and Information Security arrive or expand for similarly unromantic reasons. A growing company has accumulated things worth protecting, and protecting them requires a degree of consistency that was not necessary when twelve people could resolve most problems by shouting across the office. Customers now expect promises to be kept, regulators may have developed an interest, investors would rather not discover large surprises in the accounts, and employees themselves quite reasonably expect decisions affecting them not to depend entirely on whether the founder happens to like them. The informality that helped the company move quickly can become remarkably unattractive when you are the person on the receiving end of someone else's questionable judgement.

And there will be questionable judgement, which is where the romantic version of entrepreneurial freedom tends to collide with the reality of employing large numbers of human beings. Give people discretion and most will probably use it reasonably well, some will use it exceptionally well, and eventually somebody will make a decision so baffling that leadership briefly wonders whether allowing anybody to make decisions was a mistake. One disastrous customer commitment, spectacularly poor hire, creative interpretation of an expense policy or casual approach to confidential information can achieve more for the growth of corporate governance than several years of management consulting.

The response is usually perfectly understandable. Something went wrong, leadership does not want it to happen again, so a control is introduced. Perhaps contracts above a certain value now require Legal approval, Finance introduces another spending threshold, HR becomes involved earlier in hiring, or Security decides that employees can no longer install whatever they fancy on their laptops. Nobody has declared war on entrepreneurship. Somebody has simply looked at an avoidable problem and done what competent managers are supposed to do: learn from it.

The interesting part comes years later, when the employee responsible for the original disaster may be long gone, nobody quite remembers why the rule exists, and hundreds of perfectly sensible people are still following the process created to prevent one person from doing something spectacularly unsensible. Another incident has produced another control in the meantime, another acquisition has added another policy, and another executive has quite reasonably asked how the company can make sure that never happens again. Bureaucracy, viewed from this angle, is less something an organisation consciously builds than something it accumulates, one perfectly defensible decision at a time.

Research into organisational innovation has been wrestling with versions of this problem for decades. Formalisation and centralisation are not automatically enemies of innovation, and the evidence is considerably more nuanced than the popular idea that bureaucracy simply kills creativity. Their effects depend on the organisation, the kind of innovation and the circumstances in which it is happening.[1] That qualification matters because the interesting problem is not that successful companies become stupid. Quite the opposite. Much of the bureaucracy they accumulate is the consequence of intelligent people solving real problems, occasionally caused by considerably less intelligent people.

Think like an entrepreneur

Against this background, the messages coming from the other side of the organisation become increasingly interesting. Employees are encouraged to take ownership, challenge convention, move faster, think differently and, inevitably, think like entrepreneurs. There will be leadership presentations about innovation and perhaps a corporate value or two celebrating courage and calculated risk-taking. Someone may even invoke the Silicon Valley spirit of moving fast and breaking things, although in most mature organisations the acceptable list of things one may actually break is probably quite short.

There is nothing inherently ridiculous about wanting employees to think entrepreneurially, but entrepreneurship involves rather more than enthusiasm and initiative. It involves making decisions under uncertainty, committing resources without knowing precisely what will happen, testing assumptions that may turn out to be wrong and occasionally discovering that the clever idea was not clever at all. If the outcome were already known, the customer demand established, the return calculated and the execution risk comfortably understood, we probably would not call it entrepreneurship. We would call it a plan.

James March described a version of this tension more than thirty years ago through the distinction between exploration and exploitation. Exploration involves experimentation, discovery and the pursuit of new possibilities, while exploitation involves refining and efficiently using what the organisation already knows.[2] A growing business needs both, but success inevitably gives exploitation a rather powerful argument. Once something works, there are customers expecting it to continue working, employees whose salaries depend upon it, investors expecting returns from it and managers being measured on how reliably they deliver it. March's argument is particularly relevant here because adaptive organisational processes can favour exploitation in the short term even where doing so creates longer-term problems.[2]

Predictability therefore becomes increasingly valuable. Finance wants a forecast, Sales wants a product it can confidently sell, Operations wants something it can repeatedly deliver, customers want commitments they can rely upon, and the board would quite like management to know roughly what the business will earn next year. None of this is unreasonable, but genuine entrepreneurship lives in the uncomfortable territory where some of those answers are not yet available.

The contradiction appears when an employee is encouraged to behave entrepreneurially and then discovers that the organisation would quite reasonably like a business case before releasing the money. Finance wants an expected return, Legal wants to understand the exposure, Security needs to know what systems are involved and Procurement may want to know why this particular supplier has been selected. Before long, an experiment whose purpose was to establish whether an assumption was correct is being asked to demonstrate its return on investment before anyone has been allowed to discover whether the assumption was correct in the first place.

There is nothing particularly malicious about this because everybody is doing their job. Legal manages legal risk, Finance manages financial risk, Compliance manages regulatory risk, HR manages employment risk, Security manages information risk and Procurement manages supplier risk. Everyone can perform their role perfectly well and the organisation can still arrive at the slightly awkward question of whether anybody is left with permission to take a risk. There may, of course, be a steering committee available to investigate.

Take risks. Just don't be wrong.

That permission becomes much easier to see when something goes wrong, because the same organisation can react very differently to two surprisingly similar decisions. Someone spots an opportunity, cuts through a few procedural niceties, makes a judgement call and produces an exceptional result. The story afterwards is about initiative. They challenged convention, acted like an owner and refused to allow bureaucracy to stand in the way of the customer. Their behaviour may even appear in a leadership presentation as an example of the entrepreneurial culture the organisation would like to see more often.

Let the same person make an equally considered judgement call and have it end badly, however, and the procedural niceties become rather less peripheral. Now we want to know who approved it, why it was not escalated, why Legal was not consulted and why it was not in the budget. Somewhere along the way the organisation has moved from admiring the person's willingness to circumvent unnecessary process to reconstructing precisely which unnecessary process might have prevented the problem.

There is a well-established behavioural effect behind this. Jonathan Baron and John Hershey demonstrated what became known as outcome bias: people judge decisions and decision-makers more favourably when the outcome is favourable, even when the information available at the time the decision was made is held constant.[3] Inside a company, that bias has consequences beyond the person whose decision is being examined because everyone else gets to watch what happens next.

This is where the usual corporate language around risk becomes slightly hollow. If an employee takes a sensible risk and succeeds, celebrating them is easy. The real test of an entrepreneurial culture is what happens when a sensible risk produces a bad outcome. If the organisation looks at what was known when the decision was made, considers whether the judgement was reasonable and asks what has been learned, then the permission to take risks probably means something. If it works backwards from the failure until it identifies the approval that should have been sought, the message is equally clear.

"Take risks, but don't be wrong" is not a particularly useful definition of entrepreneurship because risk rather inconveniently includes the possibility of being wrong. Employees are perfectly capable of understanding the personal consequences of that distinction. A successful unconventional decision may produce recognition, a larger bonus or promotion; an unsuccessful one can become a career-limiting anecdote. Following the approved process offers a different form of protection because even when the outcome is poor, everyone involved can demonstrate that the correct people were consulted, the correct boxes ticked and, ideally, enough people were copied on the emails.

Nobody needs to send an email instructing employees to become more cautious. They learn by observation, watching what happens to colleagues who stick their necks out, noticing which failures are treated as learning and which require someone to blame, and discovering how much informal approval is wise even when formal approval is supposedly unnecessary. Eventually the employee who has been repeatedly told to think like an owner starts thinking rather more like an employee, which seems an entirely rational response to receiving an employee's authority, an employee's upside and an employee's ability to be fired.

Surely guardrails solve this

The obvious response is to give people guardrails rather than subject every decision to an approval process. Define the risks the organisation genuinely cannot tolerate, make the boundaries clear and give people discretion inside them. It is an appealing idea because good guardrails should remove the need for constant permission-seeking, allowing people to move quickly without pretending that a multinational corporation can operate like six friends building something in a garage.

The difficulty is that guardrails are considerably easier to describe than design. Give someone authority to spend €50,000 on an experiment and the financial boundary appears clear enough, until the experiment touches a strategically important customer, creates a contractual precedent, exposes data, changes an employment practice or creates €2 million of downstream consequences. The €50,000 guardrail starts acquiring qualifications, the qualifications acquire exceptions, the exceptions need interpretation and eventually someone decides a committee would be helpful. Given sufficient time and organisational creativity, we have successfully reinvented the approval process while continuing to call it empowerment.

There is also the awkward matter of human judgement. Give ten people exactly the same discretion and they will not use it in exactly the same way. Some will be cautious, some ambitious, some exceptionally perceptive and eventually someone will do something so intellectually adventurous that leadership begins questioning the wisdom of the entire experiment. The problem is not necessarily that people cannot be trusted; it is that judgement is unevenly distributed, while corporate policy rather stubbornly prefers to be evenly distributed.

A founder can comfortably say that Sarah is allowed to make the decision because Sarah has repeatedly demonstrated that she knows what she is doing. At scale, that becomes harder. Why can Sarah do it when somebody at the same level in another division cannot? How is the difference documented? Is it fair? Is it defensible? What happens when Sarah leaves? Before long, somebody will quite reasonably suggest a competency framework to establish who qualifies as Sarah. Standardisation starts looking rather attractive because processes scale much more easily than personal trust.

This is one reason I am increasingly sceptical of simple prescriptions about restoring entrepreneurial culture. Telling a 30,000-person company to trust its people more ignores the perfectly legitimate reason it stopped relying entirely on trust in the first place. The interesting problem is not how to eliminate bureaucracy, but how to stop an organisation designing its entire operating model around the worst judgement somebody once exercised.

What happened to the entrepreneur?

The startup had a much easier answer because uncertainty was part of its reason for existing. It did not yet know precisely which customers would buy, which proposition would work, what the eventual business model would look like or even whether the company would survive. Experimentation was not an innovation initiative running alongside the business. Experimentation was the business.

Success changes that relationship because the company now knows considerably more about what works, and there is real value in doing those things consistently. The entrepreneurial founder who once changed direction after a conversation with a customer may now have thousands of customers who would be understandably irritated if the company changed direction every Thursday afternoon. What looks like entrepreneurial agility at twenty employees can look remarkably like organisational chaos at twenty thousand, which is why I am not convinced companies should aspire to retain their startup mentality forever.

Some of what we romanticise about startups is simply what organisations look like before they have accumulated enough customers, employees, obligations and experience to require better ways of operating. Not every missing process is evidence of entrepreneurial genius. Sometimes nobody has got around to writing the process yet. The achievement of the successful corporation is precisely that it has learned how to turn something uncertain into something repeatable, and the more interesting question is what happens to its ability to pursue the next uncertain thing.

If the internal organisation has become increasingly good at evaluating investments through predictable returns, established customer demand and defensible business cases, genuinely uncertain innovation begins competing for resources on rather unfavourable terms. The existing business can explain what another €10 million will probably produce. The entrepreneurial idea is asking for €10 million partly because it does not yet know, at which point capitalism has developed a rather elegant alternative.

Let somebody else be wrong first

The large corporation does not necessarily need to conduct every uncertain experiment itself. Thousands of entrepreneurs, startups and venture investors are perfectly willing to do it elsewhere. Capital can fund multiple attempts at solving a problem, most can fail, some can muddle along and a small number can prove that there really is a market worth pursuing. There is research suggesting that this is more than a convenient story. Gordon Phillips and Alexei Zhdanov examined the relationship between R&D and acquisition markets and found that an active M&A market can strengthen the incentives for smaller firms to innovate. Their model and empirical evidence suggest that smaller companies may invest more aggressively in innovation when acquisition provides a route to realise its value, while larger companies can sometimes obtain access to that innovation through acquisition rather than competing directly in the same R&D race.[4]

That does not mean large companies have stopped innovating, nor does it mean organic growth has become obsolete. Both claims would be far too broad. It does, however, raise an intriguing possibility: perhaps part of the entrepreneurial experimentation that once happened inside large organisations can increasingly happen outside them, funded by investors whose model explicitly accommodates the possibility that many of those experiments will fail. The corporation can then enter the process later, when at least some of the uncertainty has been removed. The product exists, customers have bought it and the founders have demonstrated that the idea can work. Instead of funding ten uncertain experiments internally, the company can buy one of the businesses that survived them.

Seen this way, large companies may not have stopped valuing entrepreneurship at all. They may simply have discovered that entrepreneurship becomes considerably easier to value once somebody else has proved that it works. There is something deliciously circular about that: the mature corporation spends considerable effort encouraging its own employees to take calculated risks while simultaneously participating in a capital market capable of supplying innovation after much of the entrepreneurial risk has already been taken elsewhere. The startup gets to say, "F*ck it, let's try it." The corporation gets to say, "Interesting. Show us the numbers."

And then we buy them

The acquisition itself should solve the problem. The corporation gets the technology, talent, customers, intellectual property or business model it struggled to create internally, while the startup gets access to capital, distribution, enterprise customers and capabilities it could never have built as quickly on its own. There are plenty of acquisitions where exactly that happens, which is why it would be too easy to tell the familiar story of big companies buying innovative startups and immediately destroying them.

What interests me more is what happens organisationally after the champagne has been drunk and the people responsible for the acquisition have finished congratulating one another on the synergies. The acquired company now belongs to an organisation that has spent years developing controls for perfectly sensible reasons. Its employees need to enter the corporate HR framework, its systems need to meet enterprise security standards, its suppliers need to satisfy Procurement, its contracts need to satisfy Legal, Finance needs comparable reporting and management quite reasonably wants to understand what it bought and whether those synergies everyone was so excited about are actually materialising.

Every one of those requirements may be justified, just as every individual control accumulated by the parent company may originally have been justified. The difficulty is that the corporation is now applying the machinery it developed to create predictability to an organisation whose value emerged partly from its ability to operate under uncertainty. Somewhere around this point, somebody from the acquired company will probably utter the immortal words, "We used to be able to just get things done." Somebody from the parent organisation will think, perhaps quite reasonably, "Yes, and you also used to have seventy employees." Neither is necessarily wrong, and the tension between those two positions is probably more useful than pretending one side represents innovation and the other bureaucracy.

This takes us back to the question that started bothering me in the first place. What exactly are we trying to preserve when we talk about preserving startup culture? It cannot sensibly be the absence of process, because some of that process is precisely what enables the business to scale. It cannot be unlimited freedom, because the consequences of decisions change with size. It probably is not the founder's personality either, because any organisation dependent upon one person forever has created a different sort of problem. Perhaps what matters is something much less visible: permission to pursue an uncertain outcome without already knowing that it will work.

That permission exists naturally in a startup because there is no alternative. In the mature organisation it has to coexist with everything the company has learned about protecting customers, investors, data, money and reputation, plus everything HR will assure us it is doing for the employees. That is an inherently uncomfortable arrangement, and I am no longer convinced there is a clever organisational model that makes the tension disappear. Research on bureaucracy and innovation points in much the same direction. The relationship is not simply "more bureaucracy equals less innovation"; centralisation, formalisation, organisational context and the kind of innovation involved interact in more complicated ways.[1] The problem therefore cannot be solved by enthusiastically tearing up policies any more than it can be solved by writing another policy about innovation.

The company has to live with the contradiction. It needs predictability because success has given it something worth protecting, while it needs uncertainty because the future has not yet been invented. It needs people to exercise judgement while knowing perfectly well that some people will exercise terrible judgement. It needs controls that prevent unacceptable risks without turning every unusual decision into a request for permission, and it needs to learn from failure without pretending every failure was clever experimentation. Somehow it also needs to distinguish between a bad decision that got lucky and a good decision that did not, which is considerably harder than telling everyone to think like entrepreneurs.

Perhaps this is why I keep coming back to HR and Compliance, unfairly or otherwise. Their arrival does not kill entrepreneurship, but it marks something important in the life of a company. The organisation has become large and consequential enough that informal trust and individual judgement can no longer carry the whole load. From that moment, entrepreneurship is no longer simply the natural consequence of how the company operates. It is competing with another perfectly legitimate organisational ambition: predictability.

Nobody ever calls the meeting to announce the change, and nobody puts "We've had enough entrepreneurship. Let's build a bureaucracy" on the agenda. There is no ceremonial handover from the founders to Compliance, and I suspect HR would insist on a more carefully worded transition plan anyway. The company simply becomes successful enough to have something worth protecting, experiences enough painful mistakes to learn how to protect it, and gradually surrounds itself with the accumulated wisdom of everything it never wants to happen again.

The irony is that some of those protections will eventually make it harder to do the uncertain, unconventional and occasionally wrong things required to discover what comes next. Perhaps that is the real challenge of growing up as a company. The objective cannot be to remain a startup forever, because successful startups eventually have to become something else, but nor can it be to eliminate uncertainty, because a business that only knows how to repeat what already works has quietly made a rather large assumption about the future.

The challenge, then, is to become predictable enough to deserve the trust success requires without becoming so predictable that the next entrepreneur has to leave the company to get anything interesting done. And if they do leave, there is always the possibility of buying them back later.

References

[1] Damanpour, F. (1996). Bureaucracy and Innovation Revisited: Effects of Contingency Factors, Industrial Sectors, and Innovation Characteristics. The Journal of High Technology Management Research, 7(2), 149–173. The study examines centralisation and formalisation as distinct aspects of bureaucratic control and uses a multivariate meta-analytic approach to explore how contingency factors affect their relationship with innovation.
Reference 1: ScienceDirect

[2] March, J. G. (1991). Exploration and Exploitation in Organizational Learning. Organization Science, 2(1), 71–87. March examines the organisational tension between exploring new possibilities and exploiting existing knowledge, including the tendency of adaptive processes to favour exploitation in the short term.
Reference 2: INFORMS

[3] Baron, J. & Hershey, J. C. (1988). Outcome Bias in Decision Evaluation. Journal of Personality and Social Psychology, 54(4), 569–579. Across five studies, participants evaluated decisions and decision-makers more favourably following good outcomes despite having the same information that had been available when the original decision was made.
Reference 3: PubMed

[4] Phillips, G. M. & Zhdanov, A. (2013). R&D and the Incentives from Merger and Acquisition Activity. The Review of Financial Studies, 26(1), 34–78. Their model and empirical tests examine how an active acquisition market affects R&D incentives, finding particularly strong effects among smaller firms and showing how larger firms can gain access to innovation through acquisition.
Reference 4: Oxford Academic

Contact

Reach out for tailored support across services procurement, external workforce strategies, workflow optimisation and SaaS evaluation.

© 2026. All rights reserved.