[morningcovfefe]

Vision is obviously necessary - but a big problem folks run into is failing to recognize when something is not working. A/B testing is one way to keep yourself from falling into the trap of thinking 'I know better than my users what they should want'

[TylerE]

Also a good way to find a local maximum when there are much higher maxima not far removed.

[blobster]

I agree, but I think this doesn't apply to innovation. When is comes to innovation, people always want faster horses.

[rv-de]

that depends on the validity and accuracy of your test metric.