😵 From Pilot to Scale and the Innovation “Valley of Death”
Everyone knows how to run a pilot. The problem is what comes next. Between a pilot that worked and a product that scales across the entire company lies an abyss that Harvard Business Review researchers call “innovation’s valley of death” — the point where more than 70% of corporate innovations die, not for lack of technical merit, but because the organization isn’t ready to receive them.
Why the Pilot Works
Small, motivated team, dedicated sponsor, no bureaucracy, freedom to experiment. The pilot is an artificial environment for success.
Why the Valley Kills
Loss of the sponsor, budget cuts, resistance from the core business, lack of a process for scaling, a team that can't be replicated, wrong metrics.
What sets apart those who cross
Advance preparation: institutionalize before scaling, build the bridge before reaching the edge, ensure committed resources.
📊 Data that proves the valley
- 71% of Fortune 500 companies report difficulty scaling innovations beyond pilots (McKinsey Global Innovation Survey, 2023)
- 4 out of 5 digital transformation projects get stuck in pilots for more than 18 months (BCG Digital Transformation Report)
- R$ 2,3 bi and the estimated average annual cost of “eternal” pilots at large Brazilian banks (FGV EAESP, 2022)
✅ Criteria for Deciding to Scale
Scaling too early wasted billions. Scaling too late handed the market to competitors. The decision of when to scale is one of the most critical—and also one of the most poorly made—in corporate innovation. Most companies scale because of political pressure (“we need results”), not technical evidence. The result is predictable: costly failure.
Scale Readiness Framework — 4 dimensions
Market Readiness
The problem you solve is real, recurring, and backed by proven willingness to pay. It's not pilot enthusiasm — it's repeated behavior.
Economic Readiness
The unit economics work—or there’s a clear path to making them work at scale. LTV/CAC > 3x. Marginal costs decrease with volume.
Operational Readiness
Operations can handle the volume. Processes are documented, systems are load-tested, and the team is trained and not dependent on just one person.
Organizational Readiness
The company is ready to receive it. There’s an active sponsor, an approved budget, and the affected departments are aligned, not just informed.
✓ Green Lights for Scaling
- ✓Pilot users complain when the system is unavailable
- ✓Other areas request access spontaneously
- ✓Costs decrease as volume increases
- ✓The team can explain the process without the creator
- ✓Clearly better business metrics (revenue, cost, satisfaction)
✗ Red Flags — Don’t Scale Yet
- ✗The result depends on a specific hero on the team
- ✗Unit cost increases as more users join
- ✗Approvals "in principle" but without a real budget
- ✗Process documented only in the leader’s head
- ✗Pressure to scale comes from the narrative, not the data
🏗 Resources, Structure, and Teams for Scale
The team that created the pilot is rarely the team that should scale it. It’s not a flaw—it’s organizational biology. Founders need ambiguity; operators need predictability. Confusing the two roles is one of the most common and costly mistakes in corporate innovation. Jeff Bezos was one of the few who understood this: he created a separate structure for AWS and put an operator, Andy Jassy, in charge of scaling it.
The 3 Profiles Needed at Scale
Innovator (Founder)
Created, experimented, iterated. High tolerance for ambiguity. Essential for the next pilot.
Should: move on to the next H3 project
Scaler (Builder)
Turns what worked into a repeatable process. Builds teams, defines processes, and ensures SLAs.
Should: lead the scale phase
Operator (Runner)
Optimizes what has been scaled. Focuses on efficiency, metrics, and margins. Comfortable with predictability.
Should: take over after maturity
Timeline: Building the Scale Team
Month 1-2: Define the Scaler
Before allocating any resources, choose who will lead the scale-up. A builder profile, not a founder. A classic mistake: letting the creator lead the scale-up, then having them get overwhelmed by operations.
Month 2-4: Build the Multidisciplinary Core Team
The scale team needs: product/tech, operations, legal/compliance, finance, marketing/communications. Part-time is acceptable, but everyone needs a formal commitment—not “whenever they can.”
Month 4-8: Expand with Champions Across Departments
Every area that will adopt the innovation needs an internal champion — someone who understands it and advocates for it. This isn’t an evangelist: it’s someone accountable for local results. Nubank model: product squads with ambassadors at partner organizations.
Month 8+: Transfer to Operations
When the product is mature, the innovation team hands it off to a permanent product team. The innovation team moves on to the next challenge. This cycle is what distinguishes companies that innovate continuously.
💡 Jeff Bezos’s Rule for Scale
"If the team needs more than two pizzas to feed everyone, it's too big." In the initial phase, keep the team between 6-10 people. Larger teams create coordination overhead, politics, and slowness that destroy momentum. Scale the team only after the process is documented and tested—scaling people before the process means scaling chaos.
🤝 Integrate into the Core Without Killing Innovation
Integration into the core business is the most dangerous moment. When an innovation enters the company’s existing system, it faces organizational antibodies—processes, metrics, cultures, and incentives designed to optimize the current business, not nurture the new one. Clayton Christensen called this the "Innovator’s Dilemma": the same forces that make a company good in the present make it unable to embrace the future.
✓ Integration Strategies That Work
- ✓ Gradual Spin-In: innovation retains partial autonomy for 12-18 months, integrating processes one by one—finance first, then HR, then compliance.
- ✓ Separate metrics: maintain a distinct P&L for at least 2 years to avoid contamination by core optimization metrics.
- ✓ Permanent executive sponsor: don’t abandon sponsorship during integration. That’s when you need political protection most.
- ✓ Internal service-level agreement: define what the core will provide (infrastructure, legal, compliance) and what autonomy innovation will retain.
✗ Organizational Antibodies — Recognize Them
- ✗ "That's not how we do things here": cultural resistance disguised as standardization. A sign that no one has explained why the new process exists.
- ✗ Short-term metrics: applying quarterly ROI to innovations that need 2-3 years to mature. It kills them before they’re born.
- ✗ Core approval process: requiring 14 signatures for a change that took 1 day in the pilot. It stifles speed.
- ✗ Management change: a new VP takes over the area, doesn’t understand the history, and kills the project because they didn’t "invent it here."
Example: Magazine Luiza — Magalu Marketplace Integration
In 2012, Magalu Marketplace was launched as an internal pilot separate from the e-commerce core. Frederico Trajano (then VP of Digital) ensured the marketplace operated with its own metrics (GMV, active sellers, seller NPS) for 3 years before being fully integrated. Integration happened in stages: logistics first (to make use of existing distribution centers), then technology, and finally outcome metrics. The result: from R$ 400 mi in GMV in 2015 to more than R$ 60 bi in 2022, without “killing” the seller-centric culture the pilot had built.
🔁 Change Management at Scale (Kotter)
John Kotter, a Harvard Business School professor, studied more than 100 transformation processes at large corporations and published in 1996 the 8-step model that became a global reference. His central insight: most transformations fail not because of a lack of strategy or technology, but because of a lack of management of the human dimension of change. Scaling innovation without a change plan is like building a highway without marking the on-ramps and exits.
Kotter’s 8 Steps Applied to Scaling Innovation
Create Urgency
Show the threat of failing to scale. Use data: How much does each month of delay cost? Which competitor is coming up?
Form a Powerful Coalition
Recruit influencers from each affected area. Not just the formal ones — the informal ones (the people everyone listens to in the hallway) are more important.
Create a Clear Vision
It's not a slogan. It's a concrete picture of the future: "In 18 months, 100% of regional managers use the real-time decision dashboard."
Communicate the Vision
Repeat it 7 times in 7 formats. Email isn't enough. All-hands presentation + short video + FAQ + Q&A session + live demo.
Remove Obstacles
Map what blocks adoption: legacy system, approval process, wrong incentive, manager who doesn’t believe in it. Actively address each one.
Generate a Short-Term Win
Set 30-60-day milestones for each scaling wave. Celebrate publicly. This sustains energy and discredits resisters.
Consolidate and Advance
Each win paves the way for the next wave. Don’t declare victory too soon—the organizational resistance resurfaces if it isn’t anchored in processes.
Anchor It in the Culture
Change only lasts when it becomes part of the culture: in rituals, promotion processes, and the stories the company tells about itself.
💡 The Most Common Mistake with Kotter
Companies skip the initial steps (1-4) and go straight to implementation (steps 5-8). The result: silent resistance, superficial adoption, and regression after the first leadership change. Steps 1-4 create fertile soil. Without them, any seed dies.
🚫 Mistakes: Scaling Too Early or Too Late
There’s a paradox in scaling innovation: the two opposite mistakes—scaling too early and scaling too late—produce the same end result: failure. However, they do so for completely different reasons and have distinct symptoms. Understanding both mistakes and how to avoid them is what separates leaders who scale successfully from those stuck in the endless cycle of “pilots with no destination.”
⏰Scaling Too Early
"The pilot worked, let's expand it across Brazil now!"
- Symptom: Political pressure or euphoria after a positive pilot result without validating replicability.
- Consequence: Operations break down under volume. Cost per user explodes. The team is exhausted. Quality plummets. Dissatisfied users become detractors.
- Real-world example: Juicero (Silicon Valley, 2016) scaled production to 400 thousand machines before validating whether customers actually wanted the product. It lost USD 120 mi.
- Prevention: Require approval across the 4 readiness dimensions (Topic 2) before any scaling decision.
🕐Scaling Too Late
"It's not the right time yet; we need another round of testing..."
- Symptom: Perfectionism, fear of exposing work to corporate scrutiny, leadership that never feels "ready enough".
- Consequence: A competitor (internal or external) gets there first. The market window closes. The team loses motivation due to lack of impact. The budget is cut.
- Real-world example: Kodak invented the digital camera in 1975, but delayed commercialization for 20 years to protect film. It lost a USD 20 bi market.
- Prevention: Set a deadline for the scale decision. If there isn’t enough evidence after X months, kill the project—don’t postpone indefinitely.
How to Find the Right Moment: The Rule of 3 Simultaneous Signals
According to a 2022 MIT Sloan Management Review study of 200 successful corporate innovations, the ideal time to scale is when THREE signals appear simultaneously — and not before:
Market Signal
Organic demand: people outside the pilot spontaneously ask for access
Economic Signal
Marginal cost decreases: each new user costs less than the previous one
Operational Signal
Repeatable process: someone who wasn’t involved in the pilot can operate it without the founder
⚠ Trap: The "Eternal Pilot"
There’s a third mistake that combines the two: the pilot that never ends. There’s always one more validation, one more adjustment, one more stakeholder to convince. In risk-averse organizations, the endless pilot is a defense mechanism—it lets you show “we’re innovating” without the risk of scaling and failing. The cure is simple: at Gate 0, set the date when the pilot ends, either with a decision to scale or to stop. Pilots with no end date are cemeteries of good intentions.
📋 Module Summary
Next Module:
3.9 - Innovation Governance