If water is paid for on a "measured" basis (the more you use the more you pay), how much should you pay for each extra unit? The economist's answer would be the "forward-looking long-run marginal cost" of providing you with that extra unit.
The key phrase here is "long run", which may be best understood by contrasting it with its opposite, "short run". The short-run marginal cost would be the cost of providing that extra unit of water - the marginal unit - on the assumption that all the physical infrastructure (reservoirs, pumping stations, the network of distribution pipes and so on) is already in place. The long-run marginal cost does not take this infrastructure as given but includes an amount that reflects the cost of providing it. The forward-looking long-run marginal cost assumes that providing additional water may be more costly than providing the current supply if, for example, new reservoirs have to be built in less advantageous locations.
Economists set the price of an extra unit equal to the long-run marginal cost so that companies and customers alike face the correct financial incentive. The customer, in deciding whether to consume more water, pays a price that reflects the additional costs of supplying it. The company, benefiting from the extra revenue generated by higher water usage, will also make the right investment decisions.
So much for the theory: in practice it is much less easy to get agreement about what the numerical value should be. But while there is genuine room for disagreement, there is one reason why we should err on the low side when setting the unit price of water: if the price is set too high, the water companies will make undue profits from each extra unit sold. It would no longer be in their interests to encourage the rest of us to be economical with water (for example, via promotions of new more water-efficient appliances such as washing machines), since this would cause their profits to fall.