Among the people I interact with, and in the places I frequent, a question that comes up a lot is "what exactly is a minimum viable product?". Perhaps that tells you something about me, but let me tell you something more, and offer one answer to the question.
Back at the end of the 90s, between preparing for having a great bash for the Millenium and various other activities of the sort, I was also learning how to apply the software skills I had acquired to some sort of purpose which would pay my bills. I ended up, to a large degree by accident, to spend about four years of the rise and height of the dot-com boom at what I to this day consider the best possible school for creating great Internet apps for end users: the Helsinki offices of Razorfish, then-legendary marketing, technology and management consulting agency, now simply a legend.
In those days, we would at times come across a situation in which a pitch for a project, a client relationship, or a business idea consisted mostly of what we called the "Photoshop application" - a web site consisting of screenshots of something that had not been built. Coming from a developer point of view, and being rather proud of the skills we had collected, like any 20-something developer is, we saw these as something to laugh at. It's just a bunch of screens, it doesn't do anything! Anyone could come up with that!
Now, in some cases that's probably true, anyone could have come up with that. They weren't all great. Some were, and I would grow to respect the skills and effort people who took design seriously would apply to coming up with both great interaction, and beautiful looks for software. These are things not to be underestimated, because impressions matter a great deal, and nothing kills a budding consumer relationship faster than a dead-end transaction flow. There's something more than that to designing applications on the screen level, though, and listening to a recording of Bill Gross (of Idealab) talking at Stanford reminded me of one part of it.
At the end of that talk, he recalls the story of Carsdirect, of giving the founder a small budget and 90 days to prove there's a business. In other words, to find out whether anyone would buy a car from this site, without talking to the dealer. Turns out that once they got the web site up, in the first day four people did just that - buy a car. They would have to go and buy them from local dealers themselves and deliver the cars to their first four customers. However, what was NOT important to proving the business opportunity was whether they would be able to form a dealer relationship with auto makers, figure out the logistics of car delivery up front, and so forth. For the four first customers, it was enough to drive the cars from the dealer's lot up to their driveway one at a time, by the founder!
This is the MVP - the minimum viable proof of business. The front-end of a business is where value is delivered. Sometimes you can prove that with just by showing Photoshop images of the service to prospective customers. Sometimes you need to have a prototype site up that looks and feels like a real business. What you will not need is to figure out the supporting processes, back-end business logic, and a whole partner value network to prove business potential. Sure, those are things that you will need to figure out to turn a profit - but without the front-end facing the customers and acquiring sales, there's no revenue, there's no business, and there's no chance of profit, never mind how wonderful your back-end would be.
Looking back, I'm not sure I knew this back in the days. I was lucky to have people around me who did. Today, I still see a lot of people thinking of future businesses worry about the back-end processes before they've figured out whether there is a front-end business. Tackle the front-end first. Sometimes Eric Ries's "spend $100 on Google AdWords and see if you get any clicks" is enough to do that, sometimes you do need a web site resembling a real service. Do not waste your precious runway building out something to support even the first 10 customers through the entire product delivery before you have one customer, though! If you get even one customer, you're learning a ton about how your next product version is not what you though it would be.