(Original version written in 1995 while I was a member of the now-defunct Advanced Technology Group at Apple Computer, during a period of reassessing what ATG should be doing.)
An innovative product is one that makes a leap in the benefits-to-costs ratio in some area of endeavor.
Another way of putting this is that an innovation lowers the costs and/or increases the benefits of a task. A wildly successful innovation increases the benefits-to-costs ratio to such an extent that it enables you to do something it seemed you couldnt do at all before or didnt even know you wanted to do. Think of the following examples in these terms:
Costs can be in terms of
Benefits can come in terms of
Of course, a lessening of any cost can feel like a benefit, as less difficulty feels like greater ease, and greater speed requires less time.
Actually, the wildly successful innovations increase the benefits-to-costs ratio just as the lesser innovations do except that the increase is so huge that the innovation feels like a completely new thing. For example, before innovation you could
make copies photostat, mimeograph, etc. → Xerox get a message to someone hand carried → telegraph → telephone → email & voicemail make pictures painting → photography → digital imaging retrieve electronically published data ftp commands → web links
There is a special case that is worth noting. An innovation that replaces an m-by-n combinatorially exploded topology with a many-to-one-to-many star or hub topology yields a very large increase in the benefits-to-costs ratio. For example, before money, you had to seek out a trading partner who had what you wanted and wanted what you had. After money, you trade what you have for money, then trade money for what you want. Some other examples are
Some of these bring immense profit to their innovator, like railroads and Federal Express. Others spread their profit widely, like the Internet. One model for profiting from an m-by-n innovation is to develop it, more or less give it away, then collect P.R. benefits and whatever profits you can get from ancillary business, licensing, support, and upgrades, using your position of high brand recognition. Examples here are NFS and Mosaic/Netscape.
James Burkes "Connections" series on PBS makes a powerful case for how the web of innovation has been and always will be highly interconnected, how each innovation brings forth a paradigm shift which enables other innovations that were unthinkable in the previous paradigm.
Of course, people do develop innovations before there is sufficient infrastructure of other innovations to make them practical. This has traditionally been the space inhabited by advanced research departments. An excellent example is Apples QuickTime, which was a tin toy when it was first developed because hardware wasnt ready for it. Perhaps Apple was lucky that no one else came in early enough with enough talent and investment to surpass QuickTime. An example on the other side might be the GO pen-based computer, which had a very nice system, but the inadequate handwriting recognition was a show-stopper.
Innovating involves risking capital. If you undertake a project without sufficient supporting infrastructure, you dont know how long it will take, how much it will cost, how much market demand there will be or how soon, etc. Or maybe your project is primarily infrastructure for other innovations, and you dont know how long it will be before those other innovations pay for your project, if indeed there can be a payback model at all.
You could say that all innovation is successful by definition, in that any increase in a benefits-to-costs ratio is a success, but the innovator has to be careful if he/she wants to make a profit.
The above analysis suggests some criteria for evaluating a development effort: