The Talent Was Fine. The System Ate It.
Most innovation programmes fail because the company is wired to make building new things structurally impossible. The people were always good enough. The governance, the decision architecture, the immune system of the enterprise itself: that is what kills the venture, every time.
How information actually moves in your organisation
A support call comes in. The customer describes a problem that, if you traced it to its root, reveals a market opportunity worth millions. Here is what happens to that information in a typical enterprise.
The support agent logs it. A week later, it appears in a report. The report is reviewed by a manager who adds it to a backlog. The backlog is triaged in a monthly meeting. The meeting produces an action to "investigate." The investigation is assigned to a team that produces a document. The document is reviewed by a committee. The committee requests more data. Three months have passed. The original signal is now buried under four layers of translation, and the person who makes the decision about what to build has never spoken to the customer who raised it.
This is not incompetence. It is architecture. The system was designed to manage risk and ensure quality in the core business. It does that well. What it cannot do, by design, is move fast on something new.
Why the same structure that runs the company kills new ideas
Enterprise governance exists for excellent reasons. It prevents expensive mistakes, ensures regulatory compliance, protects the brand, and maintains the quality standards that customers expect. For the core business, it is essential.
For a new venture, it is a death sentence. Every governance mechanism that protects the core business adds latency to the venture. Procurement processes that take six weeks. Architecture reviews that take a month. Security assessments that take longer than the entire build. Hiring processes that mean you cannot staff the team for a quarter.
The venture does not die in a dramatic collapse. It dies by a thousand small delays, each individually reasonable, collectively fatal. By the time the team has fought its way through the internal machinery, the market has moved, the window has closed, and the budget review has arrived.
More than seventy percent of enterprise innovation programmes follow this exact trajectory. The pattern is so predictable it is almost boring.
How a startup does it differently
A startup with the same customer signal would respond in days. The person who hears the problem and the person who can build the solution are sitting at the same table. There is no report, no backlog, no committee. There is a conversation, a decision, and something in production by the end of the week.
The startup has no governance, no procurement process, no architecture review board. It also has no revenue, no brand, no distribution, and no infrastructure. It has one thing: speed. And that one thing is frequently enough to beat an enterprise with a hundred times the resource.
The question is not whether startups are better than enterprises. They are not. The question is whether an enterprise can create a structure that captures startup speed while keeping enterprise resources. The answer is yes, but only if the structure is deliberately designed.
The Skunkworks: speed on someone else's budget
The Skunkworks Protocol creates two things inside your enterprise: an airgap and a tunnel.
The Airgap
A genuine structural boundary between the venture and the enterprise. Not cultural. Not "permission to move fast." Structural: separate budget, separate authority, separate governance, separate hiring. The airgap prevents the enterprise immune system from reaching the venture.
The Tunnel
A controlled connection between the venture and the enterprise that allows specific resources, brand, data, distribution, and infrastructure, to flow to the venture without the governance flowing with them. The tunnel is narrow and deliberately designed. Only what helps gets through.
The combination gives the venture startup speed with enterprise resources. That is the structural advantage. A startup cannot match the resources. An enterprise cannot match the speed. A properly designed skunkworks gets both.
Three stages from experiment to verdict
Phase 1: The Contained Experiment
Months 1–6A small team, typically four to eight people, operating behind a genuine structural boundary. They have a single hypothesis, a ring-fenced budget, and permission to fail. The parent company provides infrastructure, brand, and access. The venture provides speed and learning.
Phase 2: Scaled Validation
Months 6–12The hypothesis has signal. Now test it at a scale that matters. The team grows modestly, but the boundary stays. This is the stage where enterprise gravity is strongest: every instinct in the organisation will try to bring the venture back inside the governance framework. Resist.
Phase 3: Integration or Independence
Month 12–18+The venture either stands on its own economics and earns its independence, or the learning is absorbed into the parent company. The one outcome that must be prevented at all costs is the permanent middle ground: a "venture" that is neither independent nor integrated, consuming budget and attention indefinitely.
Every instinct you have is wrong here
The skills and instincts that made you successful in the enterprise will actively undermine the venture. This is not a personality flaw. It is a structural mismatch between what the core business rewards and what a venture requires.
Instinct: Apply proven governance
Enterprise governance was designed to protect the core business. Applied to a venture, it doesn't protect anything. It just prevents the venture from moving fast enough to find product-market fit before the patience runs out.
Instinct: Staff it with your best people
Your best operators are optimised for running the existing business. They are brilliant at what they do, and what they do is completely wrong for a venture. You need people who are comfortable shipping things that are not ready, pivoting weekly, and operating without a plan.
Instinct: Measure it like the core business
Revenue targets in month three guarantee the venture will chase short-term wins instead of testing the hypothesis. The metrics that matter in a venture are learning velocity, experiment throughput, and signal clarity. None of these appear on a standard P&L.
Instinct: Keep it close for oversight
The closer the venture is to the parent company, the more it behaves like the parent company. Proximity is not oversight; it is gravity. The boundary exists precisely to prevent the parent company's immune system from killing the thing before it has a chance to prove itself.
Six conditions that are non-negotiable
Protected budget: 18–24 months
Not allocated quarterly. Not subject to reallocation. A specific number that is committed before the venture starts and cannot be raided when the core business needs to "tighten up." If the budget is negotiable, the venture is dead.
Executive sponsor
A single person at board level whose job is to keep the rest of the organisation from interfering. Not a steering committee. Not a governance board. One person with the authority to say "leave them alone" and the nerve to actually say it when the pressure comes.
Written product-market fit criteria
Before the venture starts, agree what "working" looks like. A specific hypothesis and the evidence threshold that confirms or denies it. Not revenue. Not headcount. Not "strategic alignment." A falsifiable statement and the metrics that test it.
Independent hiring
The venture hires its own people using its own criteria. Not the corporate HR process. Not the existing grading structure. Not the same interview loop. The people who run the venture choose who works in it. This is non-negotiable.
No enterprise governance
No quarterly business reviews. No stage-gate process. No mandatory architecture review. No security assessment that takes longer than the entire product build. The venture operates under its own governance regime, designed for speed, not compliance.
Minimum 18-month commitment
Innovation is not a quarter. It is not a half-year review cycle. The minimum commitment is eighteen months, and anyone who tells you otherwise is selling you a workshop. If you are not prepared to commit that timeline, save the money and put it into the core business.
The numbers your board needs to see
The case for a skunkworks is not ideological. It is financial. The alternative, continuing to run innovation through enterprise governance, has a known failure rate and a known cost. The skunkworks model does not guarantee success. It guarantees that failure, if it comes, will be fast, cheap, and informative rather than slow, expensive, and inconclusive.
>70%
of innovation initiatives fail to meet objectives
McKinsey
S&P 500
average company lifespan halved since 1960
Innosight
£2.5M
average sunk cost of a failed innovation programme
Industry estimate
What we actually do
Structural Design Sprint
We map the decision architecture of your organisation: how information actually moves, where authority pools, where the gap between signal and action makes new ventures structurally impossible. Then we design the boundary, governance model, and success framework for the venture.
Founding Team Selection
Hiring for a skunkworks is different from hiring for the enterprise. We advise on team shape, seniority mix, and the critical profile of the venture lead. Getting the founding team wrong is the single most expensive mistake, and it happens in roughly eighty percent of enterprise ventures.
Sponsor Coaching
The executive sponsor role is the most underestimated factor in venture success. We provide ongoing coaching, specifically focused on the moments where political pressure will tempt the sponsor to compromise the boundary. Those moments are predictable. We prepare for them.
PMF Review
Structured assessment against the pre-agreed product-market fit criteria. Honest signal reading with a clear recommendation: continue, pivot, or stop. No sunk-cost reasoning. No face-saving presentations. Just the evidence and the verdict.
Self-Diagnosis
Eight questions. Answer with the organisation as it is, not as the strategy document says it should be.
Can you name a specific unmet customer need that the venture will address, without using the words "platform," "ecosystem," or "digital transformation"?
Question 1 of 8
“The people were never the problem. The wiring was always the problem. Fix that first.”