An MVP is not the smallest product you can ship — it is the fastest way to learn whether your most important assumption is correct. It is defined by the learning goal, not the feature list.
Before full product development. Whenever you have a significant assumption about customer behaviour, willingness to pay, or product value that hasn't been validated.
- Identify your riskiest assumption — the one that, if wrong, makes everything else irrelevant
- Define the specific learning you need to validate or invalidate that assumption
- Design the minimum experiment that will produce that learning
- Build only what is needed to run the experiment
- Measure the specific metric that tells you if the assumption holds
- Decide: persevere, pivot, or stop — and document what you learned
Assumption: users want a personalised weekly playlist curated by algorithm. MVP: manually curated playlists sent to 50 Spotify employees each Monday, tracking how many songs they saved. Result: employees saved 3x as many songs as from their own playlists. The learning justified building the algorithm — Discover Weekly was validated before a single line of recommendation code was written.
Please contact the author for more information on these examples at linkedin.com/in/kshitijrege
- Building an MMP (Minimum Marketable Product) and calling it an MVP — these are different things
- Using MVP to justify shipping low-quality work to real customers
- Not defining the learning goal before building — an MVP without a hypothesis is just a small product
- The Lean Startup — Eric Ries
- Testing Business Ideas — David Bland