This message was deleted.

Hi Babak, this is a problem with our metric preview, but it should be computed correctly in the context of the experiment. You should see a note in the metric preview that we don't join to the denominator and are just showing you a time series of the numerator.

Thank you Luke!