Few concepts have had as much impact on the world of startups and innovation as the MVP (Minimum Viable Product), popularized by Eric Ries. The idea was simple and powerful: build the smallest possible version of a product that allows you to generate validated learning with real customers.
The problem is that, over time, many teams began to misunderstand the concept. Instead of creating the minimum needed to learn, they started launching incomplete, poorly finished products that offered little value to users.
On this point, Jason Fried, founder of 37signals, is one of the critics of the MVP concept. He argues that the goal should not be to build a minimal version, but the best possible first version within the existing constraints, as long as it is simple, lean, and genuinely useful to customers. Rather than aiming for the minimum viable product, he advocates delivering the best product possible with the resources available.
At the same time, another idea began to gain traction. Some teams realized that, in many cases, it was possible to learn something relevant even before building a product. Instead of investing time and resources in developing an MVP, they began validating their hypotheses through simpler and less expensive experiments.
This shift in perspective led many companies to question not only what should be built, but also whether anything needed to be built in order to learn.
One of the best-known examples is Dropbox, which initially used a video demonstrating how its solution would work before the product was ready. The goal was to find out whether people understood the value of the idea and were interested in using it. The strong interest generated by the video validated demand and provided enough evidence to justify developing the platform.
This kind of learning led many companies to realize that the first question is often not how do we build it?, but rather is it worth building?
MLP: when learning is not enough and you need to delight
It was in this context that approaches emerged to complement, refine, or even challenge some aspects of the MVP. One of the best-known evolutions is the MLP, or Minimum Lovable Product.
While the MVP seeks to answer the question “What is the minimum I need to build in order to learn?”, the MLP asks “What is the minimum I need to build in order to delight?” The difference may seem subtle, but it represents an important shift in focus.
The MVP prioritizes hypothesis validation. The MLP, in turn, seeks to balance learning with user experience. The idea is that, in certain contexts, offering only the minimum functionality may not be enough to win customers or generate engagement.
Consider Airbnb. An MVP might simply allow people to list available rooms. An MLP, on the other hand, would include essential elements to build trust and satisfaction from the very first interaction, such as good photographs, clear communication, and a pleasant user experience.
The MLP philosophy is based on the idea that it is not enough for a product to work. It must also make a positive first impression. After all, learning remains important, but creating an experience people enjoy is also part of validating a value proposition.
Customer Development: understand the customer before building
Long before Lean Startup became well known, Steve Blank was already advocating an idea that would directly influence Eric Ries’s work: the need to discover and understand customers before making a major investment in development.
The central principle of Customer Development became known through a memorable phrase: “Get out of the building and talk to customers.” According to this approach, a series of well-conducted interviews can generate more learning than weeks or months spent building features.
The focus shifts away from the product and toward a deep understanding of the problems, needs, and behaviors of the people a team intends to serve. For example, a team looking to create a restaurant app might discover, after interviewing dozens of owners, that their biggest problem is not order management, as the team initially assumed, but inventory control and food waste.
This discovery, made before any development begins, can prevent wasted time building the wrong solution and direct the team’s efforts toward a problem customers genuinely consider a priority.
However, this approach has an important limitation: customers are usually better at talking about current problems than radically new solutions. In nonexistent markets or disruptive innovations, such as smartphones, social networks, or streaming platforms in their early stages, users are often unable to imagine future behaviors or express demand for something they have never experienced.
In such cases, observing behavior and running practical experiments become essential complements to interviews. In practice, understanding what people do is often just as important as listening to what they say.
Pretotyping: validate interest before validating the solution
If the MVP asks “What is the smallest product I need to build?”, Pretotyping, created by Alberto Savoia, asks an even more radical question: “Before checking whether I can build this, have I confirmed that anyone actually wants it?” The idea is to test market interest with the smallest possible investment.
Instead of developing a complete application, a team can create a waiting list page, an experimental ad, a demonstration video, or even a purchase button for a product that does not yet exist. The Dropbox example mentioned earlier illustrates this approach well. Learning happens before building.
This approach gained traction because it helps prevent one of the most common mistakes in innovation: developing something technically sophisticated to solve a problem no one considers important.
Continuous Discovery: make learning part of the routine
In recent years, one of the most influential ideas in product management has come from Teresa Torres through the Continuous Discovery approach. Traditionally, many companies followed a linear sequence: research, build, and launch. Torres proposes replacing that flow with a continuous learning process.
Instead of conducting occasional research, teams hold frequent conversations with users, regularly test hypotheses, and make decisions based on evidence gathered over time. From this perspective, discovery stops being a project phase and becomes a permanent part of product work.
A well-known example of this approach can be seen at Netflix, which runs continuous experiments with different groups of users to evaluate changes to recommendations, interfaces, and features. Instead of relying only on large research studies before each release, the company continuously collects feedback and usage data, adjusting its products incrementally based on real customer behavior.
The main contribution of Continuous Discovery is precisely to make learning an organizational habit. Instead of concentrating discovery at the beginning of projects, teams maintain an ongoing dialogue with the market, reducing risk and improving the quality of decisions throughout the product life cycle.
Lean Experimentation: the experiment over the product
Perhaps the most relevant conceptual evolution of Lean thinking itself is what many people now call Lean Experimentation. In this approach, the question is no longer “Which MVP should I build?” Instead, it becomes “Which experiment will reduce my biggest uncertainty as quickly and cheaply as possible?”
The experiment does not have to be a product. Depending on the hypothesis being investigated, it could be an interview, a prototype, a landing page, a marketing campaign, or a concierge test, a practice in which a service is performed manually when the product does not yet exist.
The important thing is not to build something. The important thing is to learn.
A classic example is Zappos. Before investing in a complex e-commerce operation for selling shoes, the company tested whether people would actually buy shoes online. Its founder photographed products from local stores, posted the images on a simple website, and, whenever a purchase was made, bought the item from the store and shipped it to the customer.
The experiment made it possible to validate demand with little investment and significantly reduced the business’s main uncertainty before building the infrastructure needed to scale the operation.
The big shift in mindset
Looking at the evolution of these approaches, it is clear that none of them eliminated the relevance of the MVP. What happened was an expansion of how we think about validation and learning.
For many years, the conversation revolved around the question: “What is the smallest product I can build?” Today, more mature teams often ask: “What is the fastest way to test a hypothesis?”
This change in perspective represents a natural evolution of the ideas put forward by Eric Ries. The goal remains validated learning. The difference is that the focus has shifted from the product to the experiment.
Ultimately, the value is not in the MVP itself, but in what it helps us discover. As product practices evolve, it becomes increasingly clear that learning quickly is just as important as, or even more important than, building quickly.
And often, the best way to learn is not to develop a new feature, but to find simple and smart ways to test an idea before investing in it.
There is also a recent factor reinforcing this shift in mindset: the growing popularity of artificial intelligence. For a long time, software development was one of the main bottlenecks to innovation. Building a prototype, testing an idea, or creating a first version required weeks or months of work. Today, AI tools can generate prototypes, interfaces, and even functional applications in a matter of hours. In this context, the challenge is no longer the ability to build, but the ability to validate. When creating became easier and less expensive, the real differentiator became identifying which problems are worth solving and which solutions genuinely create value for users.
The first table below compares the goals and characteristics of each approach. The second shows how the central question in innovation has evolved over time, shifting the focus from building a product to reducing uncertainty and learning continuously.
| Approach | Main question | Focus | Primary method | When to use it | Example |
|---|---|---|---|---|---|
| MVP (Minimum Viable Product) | What is the smallest product I can build to learn? | Validating hypotheses through a minimally functional product | Building a simplified version of the product | When it is necessary to observe real use of the solution | Early versions of digital startups |
| MLP (Minimum Lovable Product) | What is the minimum I need to build to delight the user? | Combining learning with user experience | A minimal product with special attention to perceived value | Competitive markets or those where experience has a strong impact | Airbnb investing in trust, good photos, and user experience |
| Customer Development | Am I solving an important problem for the right people? | Deep understanding of the customer | Interviews, observation, and problem validation | Before starting to develop a solution | Interviews with restaurant owners to understand their real pain points |
| Pretotyping | Is there enough interest to justify building this? | Validating demand before development | Landing pages, ads, videos, waiting lists, and interest tests | When the main uncertainty is whether demand exists | Dropbox’s early video validating interest before development |
| Continuous Discovery | What are we learning continuously about customers and opportunities? | Ongoing discovery and evidence-based decision-making | Frequent conversations with users, experiments, and behavior analysis | Throughout the product life cycle | Netflix running continuous tests with users |
| Lean Experimentation | Which experiment will reduce my main uncertainty as quickly and cheaply as possible? | Learning over building | Any experiment that can generate evidence: interviews, prototypes, concierge tests, campaigns, and more | At any stage, especially when facing significant uncertainty | Zappos validating online shoe purchases before building the full operation |
The second table shows how the questions about what to do have evolved:
| Approach | Dominant question |
|---|---|
| Customer Development | Which problem is worth solving? |
| Pretotyping | Are people actually showing interest? |
| MVP | What is the smallest product that lets me learn? |
| MLP | How can I learn without compromising the experience? |
| Continuous Discovery | How can we make learning part of the routine? |
| Lean Experimentation | Which experiment is most effective at reducing uncertainty? |
Did you enjoy the review? Get your copy now: The Lean Startup