Plain version: a loss exposure is the could-happen; a loss is the did-happen. A car dealer's inventory sitting on an open lot is a loss exposure before any storm. The drop in value after hail dents the cars is the loss. The hailstorm itself is the peril.
Definitions:
Loss exposure: a situation that presents the possibility of a loss. It exists before anything goes wrong.
Loss: the actual reduction in value after an event.
Two kinds of loss exposure:
Property loss exposure: your own property could be damaged. A kitchen fire could destroy a café's ovens, wind could blow down its sign, a power failure could spoil the food in its refrigerator.
Liability loss exposure: you could be held legally responsible for injury or damage to someone else. A customer could slip on a wet floor and sue the café.
The law of large numbers: as the number of similar exposure units grows, actual results come closer to expected results. An insurer covering 500,000 similar homes can predict its losses far better than one covering 500. The law improves prediction; it does not prevent losses, and it never promises an exact match.