A real nozzle has flow losses, so its kinetic-energy increase is smaller than the reversible adiabatic increase available between the same inlet state and outlet pressure.

The isentropic nozzle efficiency compares these two increases:

$$\eta_n=\frac{V_{2,a}^2-V_1^2}{V_{2,s}^2-V_1^2},$$

where $a$ denotes the actual outlet and $s$ the isentropic reference outlet.

If inlet velocity is small, the expression simplifies to

$$\eta_n\approx\frac{V_{2,a}^2}{V_{2,s}^2}.$$

For example, if the isentropic reference predicts $V_{2,s}=500\ \mathrm{m/s}$ but the actual outlet is $450\ \mathrm{m/s}$ with negligible inlet speed,

$$\eta_n\approx\left(\frac{450}{500}\right)^2=0.81.$$

The reference state is not an additional physical process inside the nozzle. It is a reversible benchmark used to quantify the effect of losses on the useful conversion of enthalpy into directed kinetic energy.