I just finished writing a blog post that will go live in May (I write them in advance and then schedule them, to avoid forgetting), but basically I've learned a lot from those past two years, where I made some "new experiments"
While I consider PSCD a big success on gameplay aspect (the game was fun, the only critiques on that fonrt that it got was from people who remembered the first Planet Stronghold and wanted both the old manga art and RPG gameplay) the other two experiments, QoT and Amber were... less successful
Don't get me wrong, QoT was my best selling game of last year by a wide margin. Amber seems to be doing well too even if it's clearly too early. But I'm 99% sure that's not because of the gameplay
it's because most people buy them for the dating sim aspect (and was a good idea to allow a "VN Mode"!).
Anyway for future games I'll stick to the VN/RPG formula that it's already proven to work, without any "gameplay experiments". I mean I'll do experiments but within the RPG rules, where I think I'm much less likely to make mistakes. So doesn't mean all identical games, but simply experiments about RPG gameplay (like the Cursed Lands enemy personalities thing).
And if I decide to do other gameplay experiments definitely will first do the gameplay and not write the story first.
For same reason I'm going to rethink a bit Undead Lily and definitely NOT make that game another "experiment", since I think I've done enough for a while!!