Stefan Roock

All About Agile and Lean

Lean Startup: A classification of MVPs

There are different kinds of MVPs (Minimum Viable Products) with specific consequences. For me it is useful to think along two dimensions:

  1. Coverage: The number of reached customers. (An interview is done with few people, an Ad-Words campaign may reach thousands).
  2. Product Fidelity: How similar is the MVP to the end product? (A software prototype has a higher fidelity than a paper mockup).

Here are the MVPs I am aware of:


The size of the dots is the length of the feedback cycle. An interview is done in a few minutes. A software prototype needs hours or days to program and then days or weeks to gather the feedback.

How to use it

The classification is useful to find the appropriate MVP. When teams start learning Lean Startup they tend to use high fidelity, high coverage MVPs and pay the price: The need weeks to get feedback. But in the very beginning many of our assumptions are just plain wrong and we want to learn that as fast a possible. Therefore first MVPs should be as fast as possible. Using low fidelity, low coverage MVPs from the bottom left help to achieve that. When we validated some assumptions with low fidelity, low coverage MVPs we move up and to right: higher fidelity, higher coverage.

More MVPs

I am keen to discover other MVPs (the empty space in the middle of the diagram looks like there is something missing). What are the MVPs you know of?

Published by

10 Antworten zu „Lean Startup: A classification of MVPs”.

  1. It seems to me that the arrangement of data points would not look much different if we grouped them by „effort invested“ (fidelity and speed of feedback) and „support by automated tools“ (coverage).

    Following that thought, the middle might be filled by low-effort semi-automated tools. A mockup website, maybe?

  2. Hi Stefan,

    I think something like a fake feature (or feature fake) could, on a grown up product at least, count for the medium space, e.g.

    Another point that might be important to consider, when using LSU techniques in more mature products, is to keep the user in mind. The user might expect more of a MVP than the one trying to just validate an assumption. So MVPs with the user in mind might grow larger than one likes. (see my iPad example) :-/

    Also, as terrible as it may seem, in more mature products, the space of uncertainty might be quite different and the main risk offending the user rather than finding new, unchartered territory of unfound user longing and core requirements. So, again, the drift of MVPs on average might drift more to the upper right. (But of course not always! – Occam’s razor is still a good idea!).



  3. Just a thought: On the line between paper mockups and video is room for photoshopped screenshots.

  4. from Sean Murphy ( via email:

    Great post! I don’t know that these are the best three dimensions for analysis (richness, reach, and feedback time) but it’s thought provoking and represents a significant improvement over any other taxonomy that I have seen.

    I think interview, paper sketch, and paper mockup are all interviews.
    Concierge MVP, protoype, and product could all generate revenue and are in a distinct category (e.g. Marty Cagain MVP) from MVP’s purely focused on learning or probing market requirements. Video (datasheet, case study, brochure) are examples of product descriptions that would require either a conversation, survey, or embedded call to action to qualify as an MVP. Adwords are part of survey category (effectively a one button survey) could include on-line/e-mailed survey, self-assessment quizzes, and similar multiple choice or fixed answer probes.

    I think there is also a split or spectrum between trying to understand the customer’s problem and needs and getting their feedback on your product / proposed solution. Interview and Concierge MVP (also called Flinstoning or manualization) are closer than they appear on this chart. There is another MVP similar to Concierge that I have seen described as „software enabled service“ where there is purpose built technology involved but accessed by the startup to provide a specialized service. Normally concierge implies leveraging/integrating existing open source or COTS tools to emulate the results the software will deliver.

    See also for a blog on some other potential MVP’s including a list from Albert Savoia’s „Pretotyping“

  5. from Hashim Warren:
    I like the rip off MVP. Rip off a successful product to get feedback, then pivot in a different direction. This is how Burbn went from a Foursquare ripoff to a photo sharing innovation

  6. […] Roock: “Lean Startup: a classification of MVPs“, Blogpost, […]

  7. […] Article inspiré du post de Stephan Rook : Lean Startup: A classification of MVPs.  […]

  8. […] and Stefan Roock’s ( posts to provide a mapping of the different MVP an entrepreneur can use to validate a […]

  9. […] might influence your results and try to adjust for that. An MVP is developed along two criteria; fidelity and coverage. Fidelity describes how close your MVP is to your original completed concept. So a high fidelity […]

Kommentar verfassen

Trage deine Daten unten ein oder klicke ein Icon um dich einzuloggen:

Du kommentierst mit deinem Abmelden /  Ändern )


Du kommentierst mit deinem Facebook-Konto. Abmelden /  Ändern )

Verbinde mit %s

%d Bloggern gefällt das: