Scaling Intrapreneurial Success
Also known as:
Moving from proving a concept in a bounded experiment to embedding it as organisational capability — the hard transition from innovator to institution-builder that most intrapreneurial efforts fail to complete.
Moving from proving a concept in a bounded experiment to embedding it as organisational capability—the hard transition from innovator to institution-builder that most intrapreneurial efforts fail to complete.
[!NOTE] Confidence Rating: ★★★ (Established) This pattern draws on Intrapreneurship / Scaling.
Section 1: Context
Most organisations harbour isolated pockets of vitality—teams that have cracked a problem, built something customers want, or discovered a new way of working. These intrapreneurial experiments live in protected spaces: a skunkworks team with budget discretion, a pilot programme with executive sponsorship, a movement cell operating outside formal hierarchy. The system around them is often rigid, resource-constrained, or structured around older value-creation logics. The organisation itself is not yet structured to absorb, support, or scale what has been proven. This is the threshold moment: the innovation has survived the hard part (building proof), and now faces an entirely different kind of death—integration without dilution, or absorption without soul. In corporate contexts, this appears as the pilot that never becomes standard. In government, it’s the brilliant local programme that cannot replicate across regions. In activist movements, it’s the winning tactic that cannot spread beyond its originating cell. In tech, it’s the product that gained traction in a niche but cannot become platform infrastructure. The organisation has not yet learned to be the container for this new capability.
Section 2: Problem
The core conflict is Scaling vs. Success.
The intrapreneurial effort succeeds precisely because it is small, protected, and able to move fast. It has autonomy, lean structure, direct feedback loops, and permission to fail. But these conditions are fragile and finite. The moment you try to replicate or institutionalise that success—to make it work for 10 teams instead of 1, across 3 regions instead of 1, with 50 people instead of 5—the very structures that enabled success become obstacles.
The innovators face a dilemma: scale and lose the vitality that made the work succeed, or protect the original model and watch it starve for lack of resources and reach. The organisation faces its own bind: it needs the capability embedded, but embedding means bureaucracy, control, standardisation, and slowdown. Scaling also demands institutional attention (governance, resource allocation, capability-building) that diverts energy from the original work.
Worse, success attracts politics. If the intrapreneurial effort works, it becomes visible, it accumulates power, and it threatens existing departments. Institutional actors who benefited from the old way—or who fear displacement—can quietly sabotage scaling by withholding resources, imposing reporting requirements, or capturing the model into their own logic. The pattern breaks because the system sees the new capability as a threat, not an asset.
Section 3: Solution
Therefore, architect the scaling transition by building a “membrane” between the protected innovation space and the institutional body—one that allows selective permeability of capability without loss of adaptive vitality, and that explicitly mirrors the governance structure to the work structure.
The insight from living systems is this: organisms don’t scale by making every cell identical to the centre. They scale through differentiated but coordinated replication. A forest doesn’t create one mega-tree; it grows many trees, each adapted to its microclimate, each connected through the root network and the mycorrhizal commons. The scaling mechanism here is not replication—making a thousand copies of the original model—but reproduction—seeding new capabilities that maintain the genetic code (values, principles, feedback loops) while adapting to local conditions.
This means:
First, do not move the original team into an institutional role. They become the “seed source,” not the institution. They stay lean, stay experimental, stay close to the edge. Their job becomes: holding the pattern, mentoring new seeders, curating what scales and what doesn’t.
Second, build explicit “translation layers”—people or small teams who can speak both languages: the innovation language (speed, learning, emergence) and the institutional language (governance, compliance, consistency). These are not bureaucrats; they are commons stewards who understand both ecologies.
Third, make the scaling unit the cell, not the organisation. A cell is small enough to move fast, but large enough to carry the core logic. It has a clear boundary, clear purpose, and clear relationship to other cells. Each cell is seeded from the original, trained in the same DNA, accountable for the same outcomes—but given autonomy in how to achieve them.
Fourth, establish a commons layer that connects all cells: shared learning loops, pattern libraries, resource agreements, conflict resolution, and permission to deviate. This is not central control; it’s a root network.
This pattern works because it trades perfect consistency for adaptive vitality. It accepts that the model will evolve as it spreads. It also protects the original from being destroyed by success.
Section 4: Implementation
1. Protect the seed source. The original intrapreneurial team should not become the scaling infrastructure. Instead, explicitly designate them as the “pattern holders” and give them time (at least 40% per quarter) for mentoring, pattern documentation, and ongoing experimentation. In corporate settings, this means keeping them semi-protected from core business metrics. In government, it means explicit carve-out from service delivery targets. In activism, it means the core cell remains primarily a cell, not an administration. In tech, it means the original product team stays focused on innovation, with a separate team handling enterprise adoption.
2. Hire and train translators. Identify or recruit 2–3 people per scaling wave who have credibility in both the innovation space and the institutional space. A translator in a corporate context might be someone who has both shipped products and managed compliance. In government, someone who has lived in both the community and policy offices. In activism, someone rooted in the movement who also understands legal and financial structures. In tech, a person who can speak both to product/engineering and to customer success/operations. These people become the governance architects.
3. Define the cell. Get clear on the minimum replicable unit. What is the smallest, most coherent group that can carry the model forward? Is it a team of 5–7? A working cell of 15? A hub-and-spoke with 20? Define this unit’s core purpose, decision-making boundaries, and how it relates to the next cell. This is your franchise agreement, but for commons work. In corporate: each cell is a product or service line running the new capability. In government: each cell is a regional or local office. In activism: each cell is a geographic node or constituency. In tech: each cell is a user segment or geographic market.
4. Create the learning commons. Establish a structured space for sharing—monthly calls, shared documentation systems, pattern libraries, war rooms for troubleshooting. This is not a reporting hierarchy; it’s a nervous system. The commons layer should include: a shared metrics dashboard (so cells can see what’s working elsewhere), a pattern library (coded decisions and templates), a conflict-resolution process (cells will disagree; design how to navigate it), and a resource-sharing agreement (so cells in trouble can draw on the commons reserve).
5. Codify and vary. Write down the non-negotiables: core values, core feedback loops, core quality standards. For everything else, give explicit permission to vary. In a corporate scaling: “Every cell must listen to customer feedback weekly. How you do that is your choice.” In government: “Every cell must report on outcomes monthly. How you measure and what tools you use can differ.” In activism: “Every cell must practice open decision-making. The format changes by context.” In tech: “Every cell must track user activation and retention. Your metrics dashboard can be customized.” This is the DNA + freedom combination.
6. Run scaling in waves, not floods. Launch 1–2 new cells every 6 months, not 10 cells simultaneously. Each wave teaches you what breaks, what you forgot to document, what translators need to protect. Let each wave settle before seeding the next. In corporate, this is planning new teams sequentially. In government, it’s phased rollout by region. In activism, it’s strategic expansion to allied chapters. In tech, it’s market-by-market rollout with learning loops between.
7. Design governance to match the work. The governance structure should mirror the cell structure, not the institutional org chart. If you have 5 cells, create a cell-level decision council with one delegate per cell, rotating facilitation, and clear escalation paths. Governance should be smaller and closer than the institutional governance, so decisions move faster. This prevents the institutional body from strangling the scaling effort through abstract process.
Section 5: Consequences
What flourishes:
New capability becomes embedded as organisational capacity. The model spreads into places that need it, with each iteration learning faster than the last. Organisational redundancy increases—you now have multiple centres of energy, not a single fragile node. New leaders emerge from each cell, creating a pipeline of intrapreneurial talent. The learning that happens across cells—what works in cell 3 that failed in cell 2—becomes a form of organisational intelligence that wouldn’t exist in a single unit. The commons layer creates structural interdependence, which binds the scaling effort to the larger mission without requiring external mandate.
What risks emerge:
Shallow scaling is the primary failure mode: cells replicate the form without the spirit. Without strong seed-sourcing and translation, you end up with 10 mediocre copies instead of 1 excellent original plus 5 good descendants. The commons layer itself can calcify into bureaucracy, especially if institutional forces co-opt it. Cell autonomy can fragment into fragmentation—cells stop learning from each other and pursue local optimization at commons cost. This is especially dangerous in activist contexts where cell autonomy can become cell isolation.
Given that the commons assessment scores show resilience at 3.0 (below the vital threshold) and stakeholder architecture and ownership also at 3.0, watch for: institutional actors quietly defunding cells that threaten them; translators being co-opted into institutional politics rather than staying neutral; and the original seed source being asked to do too much mentoring and losing capacity for ongoing innovation. If resilience stays low, the scaling effort will be brittle—it will work until one key person leaves or one political shift happens, then crack. Strengthen resilience by building redundancy into the translator layer (don’t let one person be the bridge) and by hardening the commons agreements into structural form (contracts, governance charters, not just good intentions).
Section 6: Known Uses
Spotify’s Squad Model (2012–present): Spotify needed to scale engineering across teams without creating monolithic bureaucracy. They created “squads” (small teams with full autonomy), “chapters” (functional groups across squads), and “tribes” (clusters of squads). The model explicitly protected squad autonomy while creating learning commons through chapters. Translators were “chapter leads” who had credibility with both squad engineers and leadership. The key move: treating the squad as the replicable unit, not the entire model. Result: by 2015, Spotify scaled to 30+ squads across multiple geographies while maintaining the speed and learning velocity of their original engineering culture. The model held because it didn’t require all squads to be identical—they could vary tools and practices while maintaining shared principles.
The Brazilian Landless Workers’ Movement (MST) cell replication (1980s–2000s): MST began as a single occupation—a radical act of land seizure in southern Brazil. Rather than trying to scale “the occupation” nationally (which would have been impossible), MST developed a cellular structure: small groups of 20–50 families, each rooted in a local geography, each trained in the same movement principles and direct action tactics, connected through a commons layer of regional coordinators and a national leadership that held strategy, not operations. Each cell had autonomy in choosing which land to occupy, which tactics to use, but all cells stayed aligned on the larger vision. The translators were regional coordinators—movement veterans who understood both the local context and the national struggle. By 2000, MST had created hundreds of cells across Brazil. The model scaled because it treated the cell (not the national movement) as the replicable unit, and because the commons layer was organic (rooted in shared struggle) rather than administrative.
The UK’s Behavioural Insights Team (2010–present): BIT started as a small experimental unit inside the Cabinet Office, running small behavioural trials on public services (reducing energy use, increasing tax compliance). When trials worked, BIT faced the scaling problem: how to move pilots into institutional service delivery without killing the experimental mindset. They solved this through three moves: (1) staying small and protected themselves, becoming the “seed source”; (2) training and placing “nudge units” in individual government departments; (3) creating a commons layer of shared methodology, training, and learning calls across nudge units. Each department’s nudge unit had autonomy but shared the core BIT methodology. The translators were the departmental nudge leads—people who understood both behavioural science and bureaucratic process. Result: the model scaled to 15+ countries and dozens of government departments because it didn’t require central control. The original BIT stayed experimental; the commons kept departments connected; and local autonomy kept the work contextually grounded.
Section 7: Cognitive Era
In an age of AI and distributed intelligence, scaling intrapreneurial success gains new leverage and new risk.
New leverage: AI enables translators to operate at higher bandwidth. A single translator can now curate learning across dozens of cells using AI-assisted documentation, anomaly detection, and pattern synthesis. When cell A discovers a new approach, AI can immediately surface how it might apply to cells C, D, and F—collapsing learning lag time from months to days. The commons layer can become much thinner (fewer co-ordination meetings) because AI helps maintain coherence through shared information architecture rather than shared presence.
The cell structure itself becomes easier to prototype and evolve. Instead of waiting 6 months to see if a new cell model works, teams can run rapid simulations, stress-test governance changes, and model downstream effects before scaling. This should compress the scaling timeline significantly.
New risks: The displacement risk increases. As AI makes scaling easier (you can replicate more cells faster), institutional pressure to shrink or eliminate the original seed source intensifies. “If AI can preserve the pattern, why do we need the original team?” This reasoning is flat wrong—AI can curate and distribute a pattern, but cannot renew it or sense when it’s dying. You need living prototypes or you get calcification.
Second, AI-mediated commons layers can create false coherence. Algorithms can make distributed cells look aligned while they’ve actually diverged in values and practice. This is especially dangerous in activist and public service contexts where alignment on mission is foundational. A commons layer held through human relationship (monthly calls, shared meals, conflict resolution rituals) forces you to notice and navigate divergence. An AI dashboard can hide it until it’s catastrophic.
Third, AI commodifies capability faster. Once a pattern is codified in AI systems, it becomes much easier for competitors (in corporate contexts) or opposing forces (in activist contexts) to replicate. The advantage of a living, human-stewarded pattern is that it carries tacit knowledge and relationship that is hard to copy. AI makes the explicit parts copyable instantly. This means the differentiation has to shift from “how we operate” to “why we operate” and “who we are”—deeper cultural and relational roots that AI cannot copy. In tech scaling especially, this means your competitive advantage can no longer rest on process—it has to rest on community and culture.
Section 8: Vitality
Signs of life:
Cells are innovating within the pattern, not just replicating it. You see variations that work—different approaches to the same problem, each contextually smart. The seed source team is actively mentoring and being changed by what they learn from cells, not just defending the original model. The commons layer is alive with real debate and learning, not just reporting. When cells face trouble, they reach out to the commons for help and get real support, not process. New cells can articulate why they’re doing this work, not just how—they’re not cargo-culting. Translators are visible and trusted across institutional and innovation boundaries; people seek them out when conflicts arise.
Signs of decay:
Cells are indistinguishable from each other—the model has become so standardised that local adaptation has stopped. The original seed source team is burned out from mentoring and has no time for innovation; they’ve become administration. The commons layer has become a reporting requirement; cells dread the meetings. When cells fail, they’re quietly shut down rather than becoming learning events. New cells struggle to articulate the why; they can only recite the what. Translators are captured by the institution—they’ve become bureaucrats rather than stewards. The model spreads geographically but doesn’t evolve; it’s replication, not reproduction. You start seeing institutional actors using “scaling” as a euphemism for “control”—”we need to scale best practices” really means “we need everyone to do it our way.”
When to replant:
If signs of decay outnumber signs of life, the pattern needs redesign. This usually happens around wave 4–5 of scaling, when the original innovation has become almost routine and the commons layer has started hardening into process. The moment to replant is when you notice cells have stopped genuinely learning from each other—when the commons is transmitting knowledge but not generating it. At this point, reset: send a portion of the seed source team back into deep practice (out of mentoring), create a new commons layer with different stewards, explicitly disrupt standardisation by asking cells to run local experiments. Vitality requires renewal—if the pattern is maintaining the system but not renewing it, you’re no longer scaling success; you’re managing decline.