Choose the grain before writing a calculation. A support account can submit several tickets, an order can contain several products, and a bike station can have many trips on one day. Those repeated rows are useful activity data, but they don't automatically describe distinct people or events.
These tasks mix customer anchors, comparable reporting windows, threshold segments and presentation. Some use the bike and support datasets so that the field names no longer point to a memorized formula. Reuse the pattern and adapt the grain.
No new functions are required. Read the requested boundary carefully, including whether the cutoff is inclusive and whether an open ticket belongs in an average. In the formatting task, keep the stored measure and change its display. An attractive table can still be wrong if its denominator counts the wrong thing.