I think that honestly, the thing that makes or breaks a videogame town is the writing. Night in the Woods takes place entirely in one (not that big) town, but it’s engaging throughout because its writing is so enjoyable. All NPCs are unique, and even the most minor ones have fun personalities and little arcs that you can notice over the course of the game.
Contrast this with Dragon Quest XI, where the writing is mostly just serviceable, and almost every single person in every single city is one of like 6 stock character models. I mean, the parents of one of your party members are represented by generic character models! It’s kind of depressing. The look and layouts of the towns are pretty neat, but (with the exception of the haiku-speaking villagers of Hotto), there aren’t really any interesting people in them.
I love @alfred’s idea of making you find the cleanest bathroom. I would love a game where you have to make little practical decisions and snap judgements about the towns you visit. Where should you go to get connected with the local anarchist scene, how should you ingratiate yourself with the convenience store owner so he’ll let you post up and stake out his parking lot, how should you deal with someone making a scene at a restaurant?