If your game has a great core that's about X, or X and Y, I just want to play that part. Don't tack on all kinds of extra stuff.
This applies to smaller concepts too. If you make something that's outside the norm for your game, that's great, but if it's good enough to be in your game, why not have more of it? Amplify it.
Look through your game components, for things you like, and make more variants of them.
In my gangster game, I made a "secret base" location, that you can only go to from an adjacent location. I liked the idea, so now there are many such "secret" locations.
In Ticket to Ride: Europe, players can largely plan out which train routes across Europe they can build. However, a few routes are "tunnels", which cost a random extra amount of cards. Players will attempt to build them, and sometimes fail. This is a highly random element in a game with little in-your-face randomness, and it's jarring and un-fun to have your plans disrupted in this way.
In your otherwise-strategic game, there might be a card that revolves around psychological bluffing. It's a bit odd that it's in the game. Make bluffing a real part of the game, on numerous cards, or get rid of it.
Another common example is auctions. Is your game so mediocre that we should take a break from it for a minute or two, to conduct an auction? I hate auctions in most games. Alternatively, maybe your auction is really interesting and important. Great! Then give me more of it!
The game Keyflower has two main systems — a town-building system and an auction. It's a good game, because the auction mechanic pairs well with the town-building. Modern Art is only about auctions. Some nameless Eurogames I've played were not as good. The designer just said "this game is boring and I'm a lazy designer, so please conduct an auction, to see who goes first this round. Then, let's move onto some other random mechanics."