I’m starting to think that Minimum Viable Product is the most misunderstood term in product management today. To be fair the name is rather misleading because it’s not really a ‘shippable’ product at all. Rather MVP is a strategy for rapidly testing assumptions and ideas in the real-world. Think minimum effort for maximum learning, and not just any old learning, this is learning validated by your potential customers.
Actually, it’s much harder than just launching a product with minimal functionality, because you need to invest time in understanding what to test, how to test and analysing the results. There are many different ways of doing this, but here are 2 common approaches that help to explain the concept: prototype launch or deploy now code later. You might want to read that last one again, because at first it doesn’t seem to make any sense!
Deploy now, code later: You have an idea for a new product, let’s say selling mini-houses for squirrels, but there’s no data to justify the investment or even create a vision. You could design a basic product with the minimum functionality, build your first batch and launch, but that’s a high risk strategy prone to failure. Alternatively you could use an MVP strategy – what’s the least you can do to get the maximum learning – so you create a web site that explains the proposition, but only has a ‘tell me more’ button where the customer can provide details. You then market the website and see how many people click the button. With minimum effort you can test your assumption that people want to house squirrels and get some data on potential market size. This isn’t traditional market research, it’s real-world validated learning – in the customers mind your product exists. You can then move to the next assumption, and use MVP again until you have enough data to commit to building your first real product.
Prototype launch: Sometimes an MVP approach will mean building some functionality, but again, it’s just enough to test an assumption – remember, minimum effort for maximum learning. Sometime in the late 1990’s a UK supermarket chain thought online grocery shopping might be a good idea, but there wasn’t any data in the UK or anywhere in the world really. They didn’t know how or if customers would use the service, would it replace the weekly shop, be used for a few essentials, or something completely different? That meant they didn’t know what deliver infrastructure would be required and making a mistake there could be very expensive. They used an MVP strategy, which was clever because it hadn’t been invented yet. To understand online shopping behaviour they needed to have real customers using the service. So they built a website that allowed customers to order their shopping online but that’s all they built. Online orders were manual entered into another system, printed off in Glasgow, faxed to the local store, where a shopping assistant would walk-around and fill the trolley, before it was loaded into a van and delivered. Of course they lost money on every online order, but for a minimum investment they learnt how customers used online shopping. Armed with that data they understood where online shopping fitted into their overall strategy, built a business case and designed the distribution and delivery infrastructure required to create a successful business.
These approaches are quite different, but they’re both about getting maximum learning for the minimum effort, and not about creating shippable products – validated learning, that’s what minimum viable product strategies are all about.
And if you don’t believe me then here’s the masters voice http://www.startuplessonslearned.com/2009/08/minimum-viable-product-guide.html