Hey Tubebuddy team! I'm loving that you implemented one of the A/B improvements that I suggested a couple of weeks ago (watch time per impression). I think there might be a bug in the calculation of this metric. In the attached screenshot you can see: 1. The variation has a significantly higher click-through rate 2. The variation has a slightly higher average watch time 3. The variation has a slightly lower watch time per impression How could #3 be true if #1 and #2 are also true?