MVP and experimentation in software startups: a qualitative survey

The Lean Startup methodology disseminated the concept of MVP. Since then, the term has been used in several contexts aside startups with a blur definition. Practitioners name several artifacts as MVPs, such as prototypes or initial versions of a new product, rather than an instrument for experimentation. Given the importance of experimentation to the success rate of software startups, it is essential to understand if the experimentation element is still present in practitioners' understanding of the term and how they applied MVPs. To achieve this objective, we performed a survey with practitioners and coded their answers according to aspects found in a systematic mapping study on MVP. Our results indicate that MVP is mostly associated with the ideas of a product version and customer value rather than hypothesis testing and learning processes. Additionally, those respondents that focused on the first related group of terms did not give experiments as examples of MVP. In contrast, the opposite happened to those that used the second group.

I. Introduction

The term MVP (Minimum Viable Product) was coined by Frank Robinson back in 2001¹. Nevertheless, it got popular after Eric Ries employed it in the Lean Startup (LS). In the book that proposed the methodology, he writes: "a minimum viable product (MVP) helps entrepreneurs start the process of learning as quickly as possible. It is not necessarily the smallest product imaginable, though; it is simply the fastest way to get through the Build-Measure-Learn feedback loop with the minimum effort" and "unlike a prototype or concept test, an MVP [...] goal is to test fundamental business hypotheses" [1]. Thus, an MVP is not necessarily a piece of software; it can be any experiment that helps entrepreneurs to transform their hypotheses into facts. As an example, we could say that running an interview with potential users or customers is an MVP since it gathers information to learn about that given context. Since then, the term has spread, and several definitions appeared in the literature [2]. Nevertheless, the most recurring one is the idea of a minimum set of features to deploy a product [2], which has drifted from its original objective.