Regarding the original post. Without a large advertising budget and/or a great plan, games get totally buried. Conversely, bad games can get good exposure (sales) with good advertising. In principle, any half decent game can have good sales with good advertising. It's not all about maximising sales, but sales give money to help improve the game, and do more marketing, so a virtuous circle effect. You have the most chance at 1st release.
I co-authored a passion project game at Uni which we eventually remastered and released on Steam a few years back. 94% positive rated at ~900 reviews, but barely broke even if you costed up the labour.
The other talking point, just for fun:
Agree food is very important. In a siege, water supply was often the issue too. So much so that often castles and settlement locations were selected for the spring/well water supply. I don't think that would add much...
...but I think regarding gameplay , 'to be really accurate the biggest change would be' ...children. Implemented by a short time-lag between spawning people and fighting age.
Then, if you have any sizable army defeat you better have good city defenses or it's game over. So you'd have more Pyrrhic Victory situations, and I guess other real tactics that currently don't work. It might make game tend toward one major decisive battle.
In code it should be ok to test: Have a Age var for human Units, and say 30 seconds to adulthood, then have a proportional reduction in attack Damage points until that age. Visualisation could just scale the model x,y,z at the same rate (0.85 to 1.0). It would be subtle, but an experienced player could visually tell his opposing army was majority youth. Increasing the value of spying and game difficultly curve.
Perhaps children could be taken away like Sheep... "The ashes were trampled into the Earth, and the blood became as snow. Who knows what they came for... weapons of steel, or murder? It was never known, for their leader rode to the south, while the children went north with the Vanir."