From a post on Anders’ site from Sept 2013, like here.
What is an MVP?
- The shortest path from an idea to validated learning
- Used to avoid success theater: pretty wireframes everyone in the room like that have no actual relation to real users’ wants and needs
What is an MVP experiment?
- Identifies the highest value user pain point that can be removed for the lowest cost (collaborate with the entire team and actual users to determine the which point to work on)
- Removes that pain point to see what can be learned
- The solution can (and probably should) be imperfect, unscalable, partially designed—you want to build the fastest possible solution so that you can learn and move forward to a better solution
- Demonstrates how the solution does (or doesn’t!) help users complete their task successfully
- Give devs get a chance to implement something and see how it relates to the rest of the system
- An MVP experiment’s success stems from what we learn as much as it does from whether the tested solution succeeds or fails
- Viability is key here, as is iteration on the experiment
- Really easy to pivot to new ideas while in this state
Agile collaboration is essential to the process
- Keep documentation as light as possible to streamline process
- Focus on the outcome not the output
MVP Design Patterns
- Painkiller: Remove a high-value, low-cost element of user’s pain
- Fast Money: Ship only what people will pay for and nothing more
- Example: a financial services company wants to launch an eCommerce with with a CMS and lots of content
- Start with the smallest possible thing that encompasses what customers are paying for: expert information
- Design a test using a simple email newsletter with an out-of-the-box payment system
- Use this to validate that the content you expect to produce is content customers expect to pay for
- Wizard of Oz: Looks real on the front end, but the backend process is manual to avoid dev time at this early stage
- Go Ugly Early: If the UX design is going to be costly and time consuming, beta test a totally undesigned version to validate the concept more cheaply and quickly
- Fake Storefront: Launch only a homepage and validate consumer demand via email sign ups etc. Forces you to consider how you would market this product and what your revenue model would be
- Concierge: You simulate your software. Great for products with processes that require many steps—you test the process and correct any problems before coding begins
Test the riskiest assumption first if possible—which thing will definitely cause the product to fail? Test that thing to validate your solution
Bio from Anders’s site: I am a User Experience Designer, Product Strategist, Entrepreneur, and Agile UX Coach. Since designing and building my first commercial website for an imaging service bureau I was working for in the mid ’90s, I have been an obsessive and passionate explorer of ways to more effectively bridge the Yes/No world of computers with the Maybe-this/Maybe-that world of people.