A disinfection dose that satisfies the works can be gone by the time water reaches the far end of the network. The residual is not a property of the treatment plant; it is a property of the journey.
Where it goes
Chlorine decays in two places: in the bulk water, and at the pipe wall. Bulk decay is mostly temperature and organic load, and it is reasonably predictable. Wall decay depends on pipe material, age and biofilm, and it varies street by street in a way no model captures from first principles.
Tank turnover is the usual culprit for the worst readings. A reservoir sized for a demand the town no longer has will stratify, and the water leaving it can be days old regardless of how much passes through the inlet.
The trade-off nobody writes down
Raising the dose at the works to fix the far end raises it everywhere, and the complaints about taste arrive from the streets nearest the plant. Boosters move the decision closer to where it matters, at the cost of assets to maintain in places nobody visits weekly.
The honest version of the argument is that you are choosing where to be wrong, and measuring is what tells you which end of the network is paying for it.