Developer time
Most of the developer time on a plan goes on things that stopped working without anybody touching the site.
An API key revoked by the provider.
A connection that dropped during a refresh and stayed down.
A service moving to a new major version and taking your integration with it.
The repair is rarely the slow part. Working out which of the moving parts changed is, and that means knowing where to look and what to rule out. It goes faster when the person doing it already knows your site.
Where I can, the fix closes the category rather than the instance. If an upstream service drops a connection now and then, the alert that comes out of it waits for a second failure, so it tells you about a real outage instead of every hiccup.
Each one is written up in your repository, with the evidence and what was ruled out, so nobody starts from zero next time.
The smaller half is answering questions. How do I change this, is this normal, can the site do that. On a Statamic maintenance plan you ask me directly. No ticket queue, no support tier, no being passed around.