Thinking about it (and without any knowledge of the equipment involved) FTTP is a bit like a mobile phone network. In both directions, for a group of users, the resource (the shared exchange optical node) and onward capacity is shared amongst the users (each with their ONT) as the system architecture defines.
Each user "requests" capacity (upstream) from the exchange node or is allocated capacity "downstream" at the exchange node and that (at the simplest level) is what the infrastructure provider knows about the utilisation for each user.
There will be a contracted upstream and downstream maximum (the speed of the service) so that will be the upper limit of what will be allocated to each user.
If, at a time, the pool of users is requesting capacity in excess of what can be provided an allocation process will limit what each user can have and the "contention" in the system starts to be visible.
Currently (for the higher bandwidth services at least) I guess that contention on the optical level in the link to the premises is rarely an issue as all users on an exchange optical port would need to be requesting bandwidth in excess of the total available.
There may well be further limitations (and contention) elsewhere in the system for each ISP on the platform.
Designing a platform such as this, each piece of equipment will make data available to be gathered and processed, the challenge is to process the massive amount of raw data to get something useful for billing, network health monitoring and general support without overwhelming those operating the network.
Also with multiple parties involved in the system, it is very important for each party to be able to prove that "my part of the system is working withing the agreed service level" when it comes to the "it's not working" bunfight!