https://www.growthbook.io/ logo
#give-feedback
Title
# give-feedback
h

handsome-zebra-46141

08/26/2022, 10:28 AM
Is there a reason that the conversion window of a Metric is defined on a per Metric basis rather than a per Experiment basis?
I ask this as we often have a very similar metric with varying conversion windows. For example: • Proportion of Active Users within 2 Days of Registration • Proportion of Active Users within 5 Days of Registration etc. This is obviously exactly the same underlying SQL and it would be nice to only have to maintain one version of this SQL and be able to express what the conversion window is when the Metric is used rather than when it is defined.
1
@future-teacher-7046
f

future-teacher-7046

08/26/2022, 10:51 AM
We've talked about adding the ability to override metric settings in an experiment (conversion window, cap, min sample size, etc.). I also like the idea of a "fact table" approach where multiple metrics can share the same SQL query.
h

handsome-zebra-46141

08/26/2022, 12:52 PM
I also like the idea of a “fact table” approach where multiple metrics can share the same SQL query.
Yeah, this would be awesome too!
I also like the idea of a “fact table” approach where multiple metrics can share the same SQL query.
Another idea here. It’d be cool to be able to perform custom Windorizing at the Fact Table column level. For example, for order values we often want to cap at the 95th percentile to avoid taking into account very unusual orders but for other columns we might want to cap the bottom and top ends only at the 99th percentile.
15 Views