December 18, 2007
Product development the cheap and cheerful way
Via Marginal Revolution, via Kottke, from LinuxWorld, (isn’t the internet a wonderful thing?) comes this sentiment:
Google is one of the few large companies that gets one fundamental rule of the Internet: Trying stuff is cheaper than deciding whether to try it . . . Don’t over-plan something. Just do it half-assed to start with, then throw more people at it to fix it if it works.
I think this is quite an important insight into the evolving nature of product development. Traditionally, companies hem and haw for a long time about launching products, partly because they are hard to come up with, partly because they do not want to waste money on something that does not pan out, and partly because they do not want to damage their reputations if the products fail.
But Google has, from the start, adopted a far more relaxed and improvisatory approach to new products. Many of them, as a Business Week article on cloud computing illustrated, emerge from the "20 per cent time" that Google engineers are supposedly allowed to spend on special projects of their own choosing. Others start off small, remain in beta for some time, and eventually gather steam.
This venture capital-style approach to new products has its risks. Some Google products publicly flop and others do not go anywhere much. But it allows the company to throw things up quickly. And it does cost less than having big layers of bureaucracy deciding what should be approved and what killed.
I have heard management consultants suggest that this "portfolio" approach to product development - essentially to try out a lot of things and see what works - is the future for all kinds of companies. The LinuxWorld post expresses its benefits neatly.
Hang on a moment… Google’s approach works wonderfully for Google, but let’s be careful about how far we generalise it.
It works for Google for three reasons:
Google’s investment in developing a prototype is small - of the order of a few engineer months. Compare this the prototype cost of a new car model.
The customer’s investment in trying the new product is small. There is next to no learning time, no money to pay and no cost of failure. The worst that can happen is that my browser crashes every half hour - so what ? I wouldn’t be so relaxed about a car or plane that crashed several times a day. Even in other parts of the software business the cost of trying something new is substantial. For example, some of my clients make banking software which involves the new customer incurring costs well into seven figures implementing the product. They have to be pretty damn sure it works to take on a project of that size.
Thirdly, Google’s reputation and public persona is such that users are very forgiving of new offerings that don’t work (especially they’re not being asked to pay for them !). I’m not sure how many other companies are that fortunate.
Posted by: Alastair Dryburgh | December 20th, 2007 at 1:29 pm | Report this comment