For decades, the realm of product development has been characterized by two distinct methodologies, each with its own philosophy and operational model. While these approaches have coexisted for an extended period, significant technological advancements, particularly in artificial intelligence, are now dramatically redefining their efficacy and the strategic imperatives for product organizations. The fundamental divergence lies between the "project model," historically focused on delivering specific outputs, and the "product model," which champions the achievement of measurable outcomes. Understanding this evolving landscape is critical for companies seeking to innovate and thrive in an increasingly dynamic market.
The Enduring Legacy of the Project Model: A Focus on Output
The project model, still prevalent across many industries even in the age of sophisticated AI, operates on a hierarchical, output-driven principle. In this framework, strategic direction often originates from senior stakeholders or executives who define a prioritized roadmap comprising features and specific projects. This top-down approach then cascades down, where a "feature team product manager" is tasked with creating detailed specifications, typically in the form of Product Requirements Documents (PRDs). Designers subsequently translate these specifications into visual designs, and engineers are then responsible for building the product precisely according to these predetermined plans.
Historically, this model was born out of manufacturing principles and sequential development methodologies like Waterfall, where predictability and control over discrete deliverables were paramount. Its appeal stemmed from a perceived clarity in scope and a structured approach to execution. However, as markets became more volatile and customer expectations more fluid, the project model began to reveal its inherent limitations. Its emphasis on delivering what was asked rather than what was needed often led to the creation of products that, while technically complete, failed to resonate with users or achieve desired business results. Industry data has consistently shown high failure rates for projects solely focused on output, with various studies indicating that a significant percentage of new features or products built under this model either go unused or do not deliver substantial value, sometimes as high as 60-80%. This inefficiency, often referred to as a "feature factory" phenomenon, meant companies were expending considerable resources to rapidly produce solutions that often missed the mark.
The Ascendancy of the Product Model: Driving Outcomes Through Discovery
In contrast, the product model pivots around the achievement of tangible outcomes. This approach empowers cross-functional product teams, led by product leaders, to identify significant problems or opportunities. Instead of being handed a solution, these teams are given a problem to solve and the autonomy to discover the most effective solution. This discovery phase is paramount, emphasizing iterative experimentation and validated learning. An empowered product team’s core mandate is to first gather evidence that a potential solution is indeed "worth building"—meaning it can deliver the necessary outcome for both the customer and the business—before committing significant resources to full-scale development.
The historical roots of the product model can be traced to the agile movement of the early 2000s, which advocated for iterative development and responsiveness to change. However, it gained significant traction and refinement through thought leaders like Marty Cagan and organizations like Silicon Valley Product Group (SVPG), who championed the concept of empowered product teams. These teams, typically comprising a product manager, a designer, and several engineers, are responsible for the entire lifecycle of a product or a specific problem area. They are not merely executors but strategic problem-solvers, continuously engaging with users, analyzing data, and iterating on potential solutions. The hallmark of this model is its relentless focus on validated learning, often involving the creation of numerous prototypes—sometimes 10 to 20 or more per week—to rapidly test hypotheses and gather user feedback. This rapid prototyping, facilitated by tools like Figma long before the advent of generative AI, has been a cornerstone of effective product discovery.
The Shifting Bottleneck: From Delivery to Discovery
A pivotal change in the product development landscape has been the dramatic reduction in the cost and complexity of product delivery. Advances in cloud computing, robust DevOps practices, mature agile methodologies, and increasingly sophisticated engineering tools have made the actual construction of features and projects significantly faster and more economical. What once took months or even years to build can now be accomplished in weeks or days. This efficiency, while beneficial, has inadvertently exposed a new bottleneck: the challenge of discovering what to build.
The product world increasingly recognizes that the project model, when turbocharged by modern delivery capabilities, merely accelerates the creation of the "feature factory." Companies can now produce more bad products, faster, than ever before. The real constraint is no longer in the building process itself but in identifying a solution that genuinely merits investment. This requires a solution that addresses a real customer need, aligns with the company’s strategic objectives, generates the desired outcome, and, crucially, offers a sufficiently compelling value proposition to motivate customers to switch from existing alternatives.
Artificial intelligence, particularly generative AI, is further exacerbating this shift. While AI significantly aids in the delivery phase by automating code generation (e.g., Claude Code, Cursor) and optimizing development processes, its contribution to the discovery phase is more nuanced and fundamentally different. In delivery, AI acts as an accelerator, making it easier and faster to build what has already been decided. In discovery, AI serves as an enhancer for ideation, analysis, and prototyping, but it doesn’t replace the critical human element of "product sense" – the deep understanding of user needs, market dynamics, and business viability. This makes the product manager’s knowledge and intuition in discovery more critical than ever.
"Build to Learn" vs. "Build to Earn": A Conceptual Framework
The distinction between discovery and delivery is elegantly captured by product coach Jeff Patton, author of User Story Mapping: Discover the Whole Story; Build the Right Product, who coined the phrase "build to learn vs. build to earn." This framework vividly describes the differing objectives and approaches within the product development lifecycle, a distinction that resonates particularly strongly with product teams navigating the complexities of the generative AI era.
In both discovery and delivery, there is an act of "building," but the purpose, tools, and techniques employed are fundamentally different. Moreover, while the concept of "testing" exists in both phases, its meaning and scope diverge significantly.
Building to Learn: The Imperative of Product Discovery
In product discovery, the primary objective is to "build to learn." This phase is dedicated to exploring and validating a viable combination of technology, functionality, user experience, and business constraints that effectively addresses the four critical risks inherent in any new product or feature:
- Value Risk: Will customers choose to use or buy this solution? Does it solve a real problem or create genuine value for them?
- Usability Risk: Can customers figure out how to use this solution effectively? Is it intuitive and user-friendly?
- Feasibility Risk: Can our engineering team build this solution with the available resources and technology? Is it technically viable?
- Viability Risk: Will this solution work for our business? Can we support it, market it, and make a profit from it?
To mitigate these risks, product teams engage in rapid, iterative prototyping. With modern prototyping tools and the acceleration provided by generative AI, creating 10-20 prototypes or prototype iterations per week has become remarkably accessible, even for individuals without extensive design or engineering backgrounds. The product manager can now directly engage in creating these exploratory prototypes. The main purpose of these prototypes is to serve as artifacts for testing. "Testing" in this context means:
- Value and Usability Testing: Engaging with potential users and customers to assess whether the prototype addresses their needs and is intuitive to use. This often involves user interviews, usability tests, and feedback sessions.
- Feasibility Testing: Collaborating with engineers to evaluate the technical challenges and practicalities of building the proposed solution.
- Viability Testing: Consulting with internal stakeholders (e.g., sales, marketing, legal, finance) to ensure the solution aligns with business objectives and operational capabilities.
Building to Earn: The Rigors of Product Delivery
Once a solution has been thoroughly validated in the discovery phase—meaning the team has compelling evidence that it is valuable, usable, feasible, and viable—the focus shifts to "building to earn." This phase is about developing a commercial-quality product that can be sold, serviced, and supported, and that customers can reliably integrate into their operations or daily lives.
The risks associated with building to earn are distinct and encompass a much broader range of technical and operational considerations. These include:
- Scale and Performance: Ensuring the product can handle anticipated user loads and perform efficiently under various conditions.
- Fault Tolerance and Reliability: Designing for resilience against failures and ensuring continuous availability.
- Accuracy and Data Integrity: Guaranteeing the precision of data processing and the integrity of information.
- Privacy and Security: Implementing robust measures to protect user data and prevent unauthorized access.
- Operations and Maintainability: Designing for ease of deployment, monitoring, and ongoing maintenance.
- Provisioning and Internationalization: Supporting diverse deployment environments and catering to global user bases.
"Testing" in the delivery phase, therefore, takes on a different meaning. It involves rigorous quality assurance, stress testing, security audits, performance benchmarking, and ensuring the product adheres to all specified requirements and functions exactly "as advertised." The goal is to build a robust, reliable, and production-ready system that generates revenue and sustained customer loyalty.
Modern Practices and Industry Implications
The current technological landscape has introduced several important practical differences in how these two models are applied:
-
Accelerated Live-Data Prototypes: While the four major types of prototypes (low-fidelity, high-fidelity, functional, live-data) remain relevant, the advent of generative AI and advanced prototyping tools has dramatically reduced the cost and time required to create "live-data prototypes." These are functional prototypes that interact with real data, allowing product teams to place working versions of their product in front of select users and collect actual usage data much earlier and more affordably. This capability is a game-changer for "build to learn," offering richer insights and more robust validation.
-
Parallel Experimentation: Historically, product discovery often involved a largely sequential iteration process. Teams would start with what they believed was the most promising approach, refine it iteratively based on feedback, and only proceed to full delivery once sufficient evidence for productization was gathered. Today, the speed afforded by generative AI-based prototyping tools enables teams to simultaneously explore multiple distinct approaches to solving a problem. Several prototypes, each representing a different hypothesis, can be created and tested in parallel. This allows for broader exploration, faster identification of promising avenues, and more efficient resource allocation, as teams can quickly narrow down to the most viable options for further sequential refinement.
This emphasis on "build to learn" has profound implications for the role of the product manager. Leading companies are now recalibrating their product management interview processes to assess candidates’ understanding and proficiency in building and testing prototypes. The skills required for a strong product manager in this new era extend beyond traditional project management or facilitation. While mastering prototyping tools and discovery techniques is the "easy part," the true challenge lies in cultivating "product sense"—the intuitive understanding necessary to interpret learnings from prototypes, make informed decisions, and effectively guide product direction. Product sense involves a blend of market insight, user empathy, business acumen, and technological understanding.
The shift towards a "builder/creator" mindset for product managers is creating a divide within the profession. Some product managers, who traditionally viewed their role as facilitators or "glue" for the team, may find themselves increasingly at risk if they do not adapt. Their historical focus on project managing and coordinating outputs no longer provides the necessary strategic value. However, for those who embrace the proactive, hands-on nature of building to learn, focusing on developing their product sense and prototyping skills, this era presents unprecedented opportunities. They are poised to become the strategic architects of innovation, directly influencing product success and driving business outcomes.
In conclusion, the product development landscape is undergoing a fundamental transformation, driven by technological advancements and an evolving understanding of market dynamics. The clear distinction between "build to learn" in discovery and "build to earn" in delivery provides a critical framework for product organizations. Companies that successfully adopt the product model, empower their teams, and cultivate product managers with strong "build to learn" capabilities and acute product sense are better positioned to navigate complexity, mitigate risk, and deliver truly impactful solutions in the golden era of skilled product leadership. This shift is not merely an operational adjustment but a strategic imperative for long-term competitiveness.
