A small observability habit that prevents teams from optimizing the wrong stage of a production request.
Average latency can improve while the users with the worst experience wait even longer.
