Conversations (1)
Can we split the 'all' category into meaningful buckets? It's currently too heterogeneous to interpret.
This matches the Q3–Q4 boundary effect we charted. (edited to fix a unit)