One problem that keeps coming up is that innovation is both an act ("my team is engaged in innovation") and an artifact ("the telephone was a great innovation in its time"). It's easy to find yourself defining very different things as "innovation."
My idea of innovation as an act is that it's very close to experimentation--it's just trying something new and different from what you've tried before. A lot of innovations actually "fail" in the sense that they don't achieve their original goals. However, many innovations that fail can be very beneficial, and even an innovation that fails may jar an organization into moving in a new and more productive direction. Remember, one original application for the telephone was as a broadcast medium to transmit opera performances over a great distance. It "failed" to fulfill that purpose but had many other useful purposes.