Real loss is physical: water that entered the system and left it through a hole. Background leakage too small to surface, reported breaks, unreported breaks running until someone notices, storage tank overflows.
How to value it
At variable production cost: chemicals, power, purchased water. Not at the retail rate. Stopping a leak does not create an invoice; it stops you paying to treat and pump water nobody receives. On most systems the variable cost is a quarter to a fifth of the retail rate, which is why the valuation choice changes a headline figure by a factor of four.
The exception is a system at its supply limit, where avoided real loss defers a capital project or a purchased-water contract. There the right price is the marginal cost of the next increment of supply, and it can exceed the retail rate. That is a case to argue explicitly, per system, not a default.
Why the reported number is soft
Most filed real-loss figures are residuals: system input minus everything else the utility could account for. Anything mis-stated elsewhere in the balance lands here. That makes real loss simultaneously the largest number in many audits and the least independently verified one.
It is also why the water balance cannot be used to check a filing. In the Texas audit worksheet real loss is the plug: water losses are computed as system input minus authorized consumption, apparent loss is computed from its own three components, and real loss is whatever remains. The identity therefore holds on every filing by construction, and any error elsewhere lands silently in this line. What can be checked is how much of the apparent-loss figure the regulator supplied as a default. That is a fact about the filing, and not about the arithmetic.
Finding it
Night-flow analysis by district metered area, acoustic survey, pressure-transient work, and district-level mass balance over long windows. All of it presupposes that the zone boundaries in the GIS match the valves in the ground, which on most systems is an assumption nobody has tested.