You should look to niche down to build a Calendly that is 10x better specific group of users - it is going to be hard to make scheduling 10x better for a general audience.
Ex: I've been pretty happy with https://zcal.co which seems to target people who are looking for more personalization of invite links (I've mostly been using it because it is cheaper than other options for all the functionality I care about).
This is definitely a common issue everyone struggles with. I've noticed a lot of it stems from constantly changing my mind and self doubt during the dev process (like you mentioned, you start off building a roguelike, switch over to a solitar card game, switch over to a isomorphic strategy game).
Something that helped me a lot is first spending more time than I think I need figuring out what I want to build on a very high and broad level, and not just diving into coding. Once I figure out what to build I need to have the discipline not to change what I have decided to build. Some implementation details might change or some game mechanics might be tweaked, but on a high level (am I building a roguelike or solitar type) can't be changed once I start.
It's going to be interesting to see what they do about their IPO, it seems like their RSU's "are set to start expiring in November 2020 and in mid-2021" [1]
Ex: I've been pretty happy with https://zcal.co which seems to target people who are looking for more personalization of invite links (I've mostly been using it because it is cheaper than other options for all the functionality I care about).