This is mostly a consequence of the complexity of all the abstraction layers between "conference call" and "physical routed and switched connection". At best you might be able to identify that the visible/audible symptom is due to packet loss, but proving a root cause of the packet loss is extremely difficult. Through some tooling, like ThousandEyes (which I work on), you might be able to identify the hop in the path that's causing that forwarding loss, but unless you have access to that device it'd be impossible to prove exactly /why/ it has forwarding loss.

Any type of problem like that ultimately becomes a "5 Whys?" kind of troubleshooting to get to a real root cause, and from a end-user device you generally don't have the necessary access or data to answer more than 2-3 layers of abstraction.