Then my first step would be to check the logs, or start logging if there are none currently. You get a nice structured logging format set up, log every request

Then my first step would be to check the logs, or start logging if there are none currently. You get a nice structured logging format set up, log every request, and feed that into something you can graph and query. Leave that on long enough (long enough being a length of time that depends on what the endpoint does and how critical it is, as well as how much traffic you see regularly) and you can figure out what's being used pretty quickly.