hello, I have question about metrics percentile capping. I noticed that when adding percentile capping to a metric it changes metrics value, but it does not change the denominator. I'd imagine it should remove capped users as well as change the metric value that gets capped. Is there an explanation for this?

I'll check with the data science team

Capping does not remove users. It is a procedure known as winsorization. It takes the outlier users and sets their values to the capped value.