Imagine you're making a rocket engine, and both fuel and combustion products have high molecular weight. A rough example is burning CO (carbon monoxide) in O2, with reasoning being "both of them can be obtained from martian athmosphere". So the fuel isn't great, and specific impulse - Isp - is low, but if you add a relatively low-molecular weight component to fuel - like water - you can hope to have greater positive effect from lowering average molecular mass of combustion products than negative effect from adding an inert component.

That's partially illustrated by nuclear hydrogen engines - there is no chemistry, and the gas coming from the nozzle is actually colder than what conventional chemical rocket engine produces. Hydrogen doesn't participates in any transformations, it's purely a working fluid, yet Isp is about twice as good as with LOX-LH2 engines.

So at least adding water to burning process might sometimes improve the result.