when should you consider a distributed tracing system?

hey folks, i've been thinking a lot about debugging complex distributed systems lately. when you've got multiple microservices and asynchronous communication, it can be a nightmare to figure out where requests are actually going and what's causing latency or errors. i've seen systems like Jaeger and Zipkin mentioned, but i'm wondering at what point does setting up a full-blown distributed tracing system become worthwhile? is it when you hit x number of services, or is it more about the complexity of the interactions? curious to hear your experiences and tipping points