Distributed tracing: Is it a must-have, or are we overthinking it?

i've been setting up observability for a new microservices project and i'm on the fence about implementing distributed tracing. on one hand, seeing the flow of requests across services seems super helpful for debugging. but, it feels like a lot of added complexity and infrastructure to manage. for those of you using it, what's your take? when does it really pay off, and when is it just adding noise