View the original community article here
Last Tested: March 2021
Looker’s percentile and median measure types employ some complex sql to perform percentile logic at the aggregate level rather than in a window function. To prevent fanout through joins, Looker needs to employ percentile_distinct, but this hasn't yet been built out for every database we support.
Docs has an up to date list of what databases this is supported for, and there is a feature request in Pendo to support this globally.
Workarounds
- Use `median` or `percentile` function in table calculations.
- Create a derived table which computes the median.
This content is subject to limited support.