Every organisation says it wants innovation, usually in roughly the same language. It wants to move faster, experiment more, create new products, enter new markets and become less constrained by the habits that made it successful in the first place. Then, almost immediately, it begins constructing the machinery that will make all of those things harder.
A steering group is formed. Governance is defined. Finance asks for a business case, security introduces a review, procurement wants approved suppliers, architecture wants standards compliance, legal needs to assess the terms and risk wants evidence that an unfamiliar idea is sufficiently safe to pursue. None of these requests is absurd. In isolation, most are entirely reasonable. That is precisely why the problem is so persistent.
Large organisations rarely become slow because somebody deliberately creates one catastrophically stupid process. They become slow because sensible people keep adding sensible controls. Each one solves a local problem, protects against a remembered failure or gives another stakeholder appropriate visibility. Over time, however, those controls accumulate into an environment where almost every meaningful action requires another decision from somebody else.
At that point, the organisation has not eliminated risk. It has simply converted risk into delay.
This is one of the central contradictions in the way companies approach innovation. They want the outcomes associated with entrepreneurial organisations while preserving the control structures of mature ones. They want experimentation, but with predictable results. They want people to take ownership, but not without approval. They want speed, but also assurance that every relevant stakeholder has been consulted before anything significant happens.
Those ambitions do not coexist comfortably.
The usual response is to treat innovation as a process problem. If innovation is not happening quickly enough, a programme is created. A leader is appointed. A lab appears. Someone defines an intake mechanism, creates a scoring model and develops a set of stages through which ideas are supposed to progress. The organisation now has visible evidence that innovation is being managed.
What it often has not done is remove a single obstacle to innovation actually taking place.
The deeper issue is usually not that organisations lack a process for making decisions. It is that they require too many decisions in the first place. Every additional approval introduces delay. Every committee creates another constituency whose concerns must be satisfied. Every governance layer creates another opportunity for uncertainty to become a request for more information.
This is why organisations can be extremely busy while remaining remarkably slow. The individual participants are working hard. Meetings are happening, documents are being produced, assessments are being completed and decisions are being escalated. From inside the machinery, all of this activity feels like progress.
From outside it, very little is moving.
The accumulation of control is particularly difficult to challenge because almost every control has a defensible history. A project failed, so a review was added. A supplier caused a problem, so procurement tightened its process. A security incident occurred, so another approval became mandatory. A budget was wasted, so finance introduced a checkpoint. A team built something inconsistent, so architecture established another standard.
Viewed individually, these interventions make sense. Viewed collectively, they can produce an organisation that is increasingly optimised for preventing identifiable failures rather than creating successful outcomes.
That distinction matters because preventing bad decisions is not the same thing as enabling good ones. A system designed to ensure that nobody can make the wrong call will eventually make it difficult for anybody to make a consequential call at all.
Large organisations also have a tendency to confuse participation with decision quality. Important decisions attract more stakeholders because the implications are significant. Then more stakeholders are included because excluding them appears risky. Before long, what began as a relatively simple choice has become a collective exercise in alignment.
Should we run a pilot? Should we test a new technology? Should we work with an unfamiliar supplier? Should we stop funding something that clearly is not working? These are often treated as questions requiring broad organisational consensus when what they really require is a clearly accountable owner.
The distinction between consensus and accountability is easily lost. Consensus feels safe because responsibility is distributed. If everybody participated in the decision, nobody can easily be blamed for the outcome. But responsibility distributed too widely is often responsibility diluted beyond usefulness.
This is one of the reasons innovation programmes become oddly ceremonial. A company can create all the visible signs of experimentation while quietly ensuring that nobody has enough authority to take a meaningful risk.
The better organisations are not necessarily less governed. They are governed more deliberately. They distinguish between decisions that genuinely require scrutiny and decisions that should simply sit within an agreed boundary. They understand that governance is not supposed to mean senior people deciding everything. Its purpose is to define where authority sits, what constraints matter and when escalation is genuinely necessary.
That requires a different question from the one most organisations ask. Rather than asking how to improve the quality or speed of every decision, leaders should ask which decisions should exist at all.
Does every experiment require a business case? Does every technology choice need a committee? Does every customer pilot need executive approval? Does every deviation from a standard create material risk? Does every investment require the same level of evidence before the organisation has had an opportunity to learn anything?
Some controls protect the organisation. Others protect the process.
The distinction is worth examining.
One useful way to think about this is reversibility. Organisations often govern decisions according to category rather than consequence. A technology purchase triggers one process. The use of customer data triggers another. External suppliers, AI, customer interaction and new infrastructure each activate their own machinery.
What gets less attention is how difficult the actual decision would be to reverse.
A pilot can be stopped. A prototype can be discarded. A supplier can be replaced. A small investment can be written off. A technical choice made within a contained environment can often be revisited. These decisions are not consequence-free, but neither are they equivalent to replacing a core platform, committing to a ten-year outsourcing contract or fundamentally changing the operating model of the business.
Treating reversible and irreversible decisions as though they deserve the same governance is not caution. It is simply poor organisational design.
The role of senior leadership in this problem is particularly uncomfortable because leaders often create the very delay they complain about. Executives demand faster execution, then ask to see more decisions. They want additional visibility, more assurance, extra checkpoints and earlier escalation. The intention is normally sensible: they want to help the organisation make better choices.
The unintended effect is to make progress increasingly dependent on senior attention.
Teams learn that consequential action requires permission. Judgement becomes escalation. Ownership becomes presentation. Leaders then become frustrated that the organisation lacks initiative.
You cannot demand entrepreneurship from people while requiring them to seek approval for every meaningful act.
Authority has to travel with accountability. Without that, accountability becomes largely performative. People are held responsible for outcomes they were never genuinely empowered to shape.
The answer is not to remove governance. That would merely exchange bureaucracy for chaos, which is not the revolutionary improvement some organisations imagine. The answer is to design governance around boundaries rather than permission.
Those boundaries might include budget limits, data classifications, security requirements, customer-impact thresholds, technical constraints, legal obligations and explicit conditions for escalation. Within them, accountable teams should be able to proceed without repeatedly asking whether they are allowed to do their jobs.
This changes the role of governance. Instead of acting as the place where decisions are made, governance defines the environment in which decisions can safely be made elsewhere.
That is harder than running a committee. It requires leaders to be explicit about risk appetite rather than hiding behind process. It requires trust. It requires accepting that delegated authority will occasionally produce a decision that a senior leader would not personally have made.
That is not a failure of delegation. It is the cost of it.
Organisations that are unwilling to pay that cost should be more careful about claiming they want speed.
When companies talk about innovation, the instinct is usually to ask what needs to be added. What programme? What platform? What capability? What methodology? What investment?
A more useful starting point is often the opposite.
What decisions can we remove?
Innovation does not require the absence of control. It requires a much better understanding of where control is useful.
If an organisation wants to move faster, it can spend years trying to accelerate every approval, committee and governance process it has accumulated.
Or it can require fewer of them.