If we want to speed up things, we’re often looking at timelines: activities spread out give a good feeling on how long things take and which parts takes the longest to process. We use this timeline to get a sense of the points we can have influence on to move faster.
However, in the brilliant book Product development flow, the author explains that we should be looking at queues in our process, instead of timelines.
Queues (or moments when things are waiting to be picked up for a next step in the process) are necessary, simply because it is too expensive to be ready for peak workloads all the time. At the same time, queues are a really good indicator where (often cheap) optimizations can be made.
It all starts with metrics. For whatever process you’re working on, do you know how many queues there are? Do you know the average time something is in a particular queue? What can you say about the cost of delay in each situation?