Hunch-to-Scale Innovation Process
Develop an unlikely hunch through nonlinear experiments at increasing scale.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 6
- Confidence
- 97%
The Hunch-to-Scale Innovation Process replaces the demand for a brilliant, finished idea with an explicit learning journey. Start from a hunch, often created by combining things that appear not to belong together. Instead of defending that combination, try to prove why it will not work and use each obstacle to reveal missing knowledge. Idea formation and execution remain coupled: experiments alter the concept, while the evolving concept determines the next experiment. The endeavor grows through progressively larger tests and continual reformulation until useful traction appears or interest and evidence justify stopping. What outsiders later call a pivot is usually the visible result of many small internal changes. The process outputs a scalable organization and validated direction, not fidelity to the initial idea.
Origin
Perez-Breva developed and taught this nonlinear approach through MIT’s Innovation Teams program before presenting it in Innovating: A Doer’s Manifesto. Extracted from Masters in Business.
Core principles
- 01A new idea begins with more unknowns than answers.
- 02Unlikely combinations reveal useful problems and opportunities.
- 03Execution changes the idea while the idea changes execution.
- 04Innovation develops nonlinearly rather than through separate ideation and execution phases.
- 05Traction should emerge from incremental reformulation, not scheduled pivots.
How to run it
- 1
Express a Hunch
Describe what seems possible or intriguing without pretending that it is a complete idea or reliable prediction.
Pro tip Use tentative language that leaves room for substantial change.
Watch out A polished pitch can make people emotionally committed to premature assumptions.
- 2
Create an Unlikely Combination
Put together two capabilities, observations, or domains that do not seem to fit. Use the mismatch to expose assumptions and overlooked possibilities.
Pro tip Draw from disciplines that normally operate in separate silos.
Watch out Absurdity alone is not useful unless it generates a testable question.
- 3
Ask Why It Fails
Identify why the proposed combination does not work at its current scale. Convert each obstacle into a learning question.
Pro tip Try to disprove the hunch rather than collecting only supportive opinions.
Watch out Do not hide assumptions merely because they make the concept easier to sell.
- 4
Build the Next Test
Assemble the parts needed to test the most important unknown. Let the act of building revise both the hunch and the execution plan.
Pro tip Choose the smallest test that creates meaningful evidence.
Watch out Separating ideation from execution produces false linearity.
- 5
Reformulate Incrementally
Use new evidence to change the problem statement, combination, or implementation without staging an arbitrary change of direction.
Pro tip Maintain a record of small changes so apparent pivots can be traced to evidence.
Watch out Do not ask how to pivot before there is data indicating what should change.
- 6
Advance Through Scale
Test the evolving endeavor with more people, resources, or demanding conditions. Continue until it gains traction, reaches its natural scale, or no longer warrants attention.
Pro tip Evaluate whether the organization can reproduce the result, not merely whether one demonstration works.
Watch out An idea itself does not scale; the organization and its capabilities must do so.
In the wild
Perez-Breva’s company focused on the actual problem—determining where a caller was during an emergency—rather than defending conventional cell-tower triangulation. Available environmental data became the input for an AI system that learned to locate callers.
→ The team produced a scalable emergency-location capability by changing the technical means while preserving the problem.
Netflix began with physical DVD rental, moved into streaming, and later produced content. Machine-learning capabilities persisted and helped the organization operate at each new scale, even as the visible business became a different company.
→ Incremental capability building enabled repeated organizational reinvention rather than one isolated pivot.
Common mistakes
Waiting for the Perfect Idea
A novel idea cannot contain all the knowledge required to make it exist in the world; insisting on perfection prevents learning.
Separating Ideas from Execution
Treating ideation and execution as consecutive phases ignores how each must continually modify the other.
Pivoting on Schedule
Forcing periodic changes of direction without evidence substitutes startup ritual for incremental problem solving.
Is it for you?
Best for
It is best for novel products, technologies, and problems whose final users, requirements, or solutions are not yet knowable.
Not ideal for
It is not ideal for routine work governed by a stable recipe and fully known requirements.
From the transcript
“All ideas are effectively born bad.”
“Execution will change your idea, and your idea will change what you execute on continuously at different scales.”
“There was no pivot.”
From the episode
Luis Perez-Breva Explains How to Be ‘Productively Wrong’