Slashdot Mirror
How Google Decides To Cancel a Project
The New York Times is running a story about the criteria involved when Google scraps one of their projects. While a project's popularity among users is important, Google also examines whether they can get enough employees interested in it, and whether it has a large enough scope — they prefer not to waste time solving minor problems. The article takes a look at the specific reasons behind the recent cancellation of several products.
"Dennis Crowley, one of two co-founders who sold Dodgeball to Google in 2005 and stayed on, said that he had trouble competing for the attention of other Google engineers to expand the service. 'If you're a product manager, you have to recruit people and their "20 percent time."' ... [Jeff Huber, the company's senior vice president of engineering] said that Google eventually concluded that Dodgeball's vision was too narrow. ... Still, Google found the concepts behind Dodgeball intriguing, and early this month, it released Google Latitude, an add-on to Google Maps that allows people to share their location with friends and family members. It's more sophisticated than Dodgeball, with automatic location tracking and more options for privacy and communication."
Oddly enough, the process isn't -too- different in smaller companies. You don't actually build the prototype, but since you can get direct access to the owner/boss, you can present your idea to him/her. If they like it, you'll get to build it.
It's medium-sized companies that have the problem. They can't afford the 'see what sticks' approach, and you can't talked to the owner/boss personally, so there's no way for this to happen without some favoritism being involved. Even companies that have 'good idea' programs rarely actually get them.
"If you make people think they're thinking, they'll love you; But if you really make them think, they'll hate you." - DM
That's different really. Google's approach is to build something and see what the Internet community likes. The small business approach is to pitch something and see what the boss/owner likes. A bit different.
But at medium to larger businesses, you do get the chance to present new ideas to higher levels of management. If you pitch an idea and it gets some attention, your group can be given funding to produce a prototype.
Still, usually no one outside the company gets to see these prototypes. There are a couple of exceptions -- in the auto industry you have events like the North American International Auto Show where prototypes are tossed at consumers and marketroids note how the media and how consumers react to the prototypes. That information fuels decisions about new models for the next few years, typically.
But Google's approach is altogether different. First off, the vast majority of their money is tied up in infrastructure, not development. Producing a new product doesn't cost nearly as much as it would a traditional software house.
Think of it this way: what's it take to produce an N-tier enterprise intranet app? An analyst/project manager, maybe a couple of page designers, a couple of domain experts, a 3-4 core software developers, a DBA and a systems/network administrator. And you can do it with half that if you have pay for a team of highly-experienced superstars.
So the reason they can afford to 'toss shit against the wall and see what sticks' is because they're already spending on the infrastructure -- that cost is known and somewhat fixed. The variable part, the new development, costs relatively very little.
My blog
This makes sense only if the new projects are, on average, equally lucrative. (Apple's iPod, for example, is highly profitable.) Otherwise, the money could be better used to generate dividends or do share buybacks. Companies can't grow bigger indefinitely. At some point, a successful company should start generating a consistent revenue stream for its owners (shareholders).