1) There has been a lot of thought that's gone into how we should solve this problem. We've played with several different concepts but have ultimately settled on one; starting this weekend, we're adding standard analytics "events" to Bestly where you can send any arbitrary event that you want to track. This will then allow you to visualize how "x experiment, variation A" affects some arbitrary event elsewhere.
2) Yes! We agree. We should be displaying both our confidence level as well as statistical significance rather than the number 100.
Also, I love feedback, so feel free to email me anytime. james [at] best.ly