“Designing MVP Experiments” – Agile Experience Design Meetup @ Pivotal Labs NYC, 04/15/13

Right off the heels of the Lean Startup Machine NYC weekend, I participated in last night’s Agile Experience Design Meetup where the topic/workshop involved running through how to test an ACTAUL user problem through forming a super low functioning, but TESTABLE MVP.  LSM began its focus at point 0 in the process where we needed to validate who the customer is, if the customer actually has a problem and what the actual problem is.  Last night we came into it assuming that there is a user that does have a problem.  Unique to a lot of Meetups I’ve been to, we spent some time running through the high level process in workshop form to get us thinking along the build, measure, learn feedback loop.  Some of the thinking is counterintuitive to how people may want to think (jumping into solutions without thoroughly measuring and learning…) so the exercise was helpful.

Interesting/Important Takeaways:

  • An MVP (Minimal Viable Product) is an experiment and should be testable. 
  • It’s very important to note that an MVP is NOT necessarily a minimum feature set- It’s really the smallest amount of work which will give you the ability to test and learn.
  • Going further with the last point, it’s commonly thought that an MVP should be a prototype (sketches, lo-fi wireframes, visual design…).  A prototype can certainly be an MVP, but we don’t want to start here for the big reason that when testing on customers, many times it will result in a false positive, as it is easy for people to see functioning software and agree with the hypothesis being tested.  It makes more sense from a research and time standpoint to begin testing with a non-prototype MVP which can include email survey blasts, sign-up boxes, forms, phone calls…
  • A good MVP does not look out of place and the design should blend with the current product style.  If an MVP’s design appears out of place, there is a greater chance for skewed data.

Process:

  • Define customer problem
  • Investigate assumptions 
  • Build a test
  • Measure customer behavior
  • Evaluate success
  • Pivot/persevere