You can now create more precise metrics for Snowflake Native Experiments. We've added two new options in the metric creation flow:

1. Windowing: Set a period of time from exposure from which to include metric events in the calculation, e.g., only include conversions within 7 days of exposure. This allows more precise measurement and eliminates bias from users that have been exposed for longer in the experiment.
2. Winsorization: eliminate outlier's effect on experiment results by capping metric event values at a percentile of the total dataset.

These settings allow for further specificity and statistical clarity of experiment results. See our documentation for windowing and winsorization to get started.