Here's the problem -- you're throwing a lot of user data away. Are you saving the mouse position every mousemove or click? Are you continually monitoring if t

Here's the problem -- you're throwing a lot of user data away. Are you saving the mouse position every mousemove or click? Are you continually monitoring if that page has focus? Are you turning on the videocam to see where they're looking at, if they haven't turned away towards something else? A/B testing could work, with enough similar page widgets and enough data, but people need to collect orders of magnitude more data.