Let’s start with a haiku:
Research a concept?
How accurate will it be?
Build something, then see.
I’ve written a couple of articles recently on the difficulties in developing successful products and innovating in large organizations. A lot of what I wrote had to do with research and ROI on initiatives. And certainly for companies such as McDonald’s or Proctor and Gamble, the cost of launching a new product is significant, and the cost of launching a failure is even greater.
But in the software world, launching a product or service is very different. One (or two or three) people can successfully create something, launch it on the web, get feedback quickly, modify the offering, get more feedback, modify the offering some more, grow the audience, get more feedback….you get the picture.
Pierre Omidyar started eBay (originally AuctionWeb) in his spare time when he was working at another software company in Silicon Valley.
And of course, Google started out as a Ph.D research project for Larry and Sergei, with the idea being that pages that have more links pointing to them should be more relevant in a given set of search results.
While these are three HUGELY successful companies, there are many other smaller ones that follow the pattern these companies had in their origins. i.e. build something simple, useful, cool or different, get people to use it, get feedback, build it out a bit more etc. etc.
There was little or no market research in the founding of these companies. What research could actually have been done? Who could have predicted the success of any of these companies very early on? No one.
On a slightly smaller scale, and a little more recent success, here’s an interesting example of a 1 (OK now it’s 2) person company, started in 2003, that is generating about $10,000,000 of revenue per year. The NYTimes article provides a bit more detail. The article title alone — From 10 hours a week, 10 million per year — says a lot. The site was started as a way for the founder, Markus Frind, to teach himself ASP.NET.
The point of all these examples, and there are many more like them in the software world, is that given the very low barrier to entry, a “good idea” can be quickly iterated on and developed and become a successful endeavour without a lot of research, and market segmentation, and problem identification, and persona development etc.
Just take your idea, write some code, put it out there and see where it takes you.
Is this a valid model for developing software, particularly in the context of the WWW? I’d like to hear your thoughts.