A big question to me is what the real meat of a West Marches campaign is supposed to be. The thing that players come to play for.
In games with complex combat system, this can be the fun of playing tactical skirmish encounters.
In games with central stories, it can be to see the characters improve and become accomplished.
In a dungeon crawler, it can be fun working on puzzles and other challenges.
For a West Marches campaign, I find the first two less suitable. Complex combat tends to lead to long fights, which means slow progress. This means that either adventures get quite long, or there is little that gets done if the adventure is kept short.
With constantly changing characters, and possibly quite large pools of characters, there is little room to express the personality of individual characters. And with adventures of limited length, also not much time to indulge in such things. I am not sure where this idea is coming from, but I also feel that such a campaign also should have real dangers of PCs getting killed. (It does not seem to be an automatic requirement based on the overall premise.) Players should assume that there is a considerable likelihood for any character to die, and be open to introduce new characters when scheduling issues make it a convenient solution. This generally doesn't work well with players getting deeply invested in their darling characters.
And I personally just never understood how puzzles can be made to be both fun and appear plausible within a believable game world. I know many people love them, but I simply don't get the appeal.
I think a kind of enjoyment that a West Marches game is very well suited to provide for players who don't make the campaign about their characters, is for the adventures to reveal a world. The structure that allows players to set their own paths and pace of exploration is uniquely suited to creating a sense of discovery. Because players know that they are not walking through a curated experience like a movie or most videogames, and that there are many things that will remain hidden from them unless they take actions to reveal them. And they will have to overcome obstacles to reveal them.
Because of this, I believe that a West Marches campaign will benefit greatly from solid and thought out worldbuilding. In a dungeon crawl focused campaign, where the obstacles are the meat of the game, you can get away with generic Fantasyland as the nondescriptive overworld where the entrances to on the fly created dungeons are located. But for a more exploration focused campaign, I think exploring rooms is much less interesting than gathering pieces to assemble a greater picture. However, the challenge in this respect is that the players would need to have a constant sense of accomplishment. A player who plays only three times over four months should still walk away with a feeling of having discovered something interesting. As do the players who play every week for three years. Perhaps one approach to this is set things up so that every site has a story that can be partly pieces together just from clues in the environment, but also interactions with the current residents. But in addition to that, each site also helps providing context for a big picture to create a long-term progression for the regular players.
In addition to sites, I believe a lot can be done in regards to worldbuilding with encounters on the way to and from the sites. I am even having a hunch that a campaign could be possible in which random wilderness encounter can be main form of content for the game. But these would need to be more than just attacks by wolves, goblins, and bandits. There would have to be a well established ecosystem of bandit gangs and goblin tribes, who are having their respective territories and ongoing conflicts, and occasionally try to use interactions with players to use them to their advantage against their rivals. To that you can add various cults, and possibly guilds, and certain wizards and druids. Gathering information, through interactions, about these factional ecosystems can also be a great source of discovery. Especially when it is connected to the histories of the sites.