Counts and durations over a window, sliced however you ask. Takes the same filters as the search, so a figure here and the rows the search returns always describe the same population. Each groupBy switch adds a dimension, so groupByAssignee and groupBySeverity together give one row per person per severity. Set an aggregation to bucket by time, which also populates openAtBucketEnd - the backlog carried at the close of each bucket, which is the number a trend line should plot.