You have a 'spine' of fibre running down a road with - lets just say - ten strands of individual fibres. At some point you need to split off one fibre, send it to (up to) 32 houses down a side road. How is the main cable physically split? Is it unsheathed and a fibre exposed?
Then you have nine strands continue down the main road. At this point there are still ten but one is unlit and redundant? At some point are there more redundant strands, or at some point does is the cable replaces with ( say) one containing four individual fibres?
I don't know why, but I need to know!
Working backwards: at the endpoint, there are CBTs with say 4 or 8 or 12 ports. They are sized to cover the number of properties served.
As I understand, generally each CBT has its own cable all the way back to the splitter node - e.g. for a 12-port CBT there will be a 12-fibre cable running straight to the splitter. In principle a splice joint *could* be put along the route, but they really prefer not to do this.
At the splitter node, the appropriate number of fibres from each CBT will be attached to a splitter (these are tiny by the way, smaller than a matchbox - with fibre pigtails coming out into separate trays). If the splitter node serves more than 32 properties then there will multiple splitters.
From there, there are cables which run back to the fibre aggregation node, lighting a strand for each splitter. Again, it's *possible* to have splices along the route, but they prefer not to do that, since each splice is potentially a weakness.
For my FTTPoD install, I was talking to the engineers about this. They said that initially they had considered running a cable in sections with 4 or 5 splices along the route - presumably that would make it easier to branch off later to serve other splitter nodes. In the end they decided to do the "traditional" approach of pulling a subduct (tube) all the way from the fibre aggregation node to my splitter node, and blowing a cable in one piece through that. More reliable.