This is a lesson I am slowly coming to learn. Truth is, you just don't know what people will find valuable or not.
You can have guesses - and if you are one of the "people," one of the tribe or niche, then your guesses are probably better than an outsider's. But you do not know what will sell and what won't.
This is a sobering - and anxiety-provoking - thought to have. There is no secret formula to knowing exactly what someone will find valuable. There are just too many variables. (This is why big companies pay big money for big data. Likewise, this is probably where AI is going to go.)
Good thing is: you now have more access to your target audience than ever before. You can connect with them, tease them, provoke them. You can test with them.
My suggestion, then, if you're looking to create things that people buy: test, test, test. Test fast. Create a Facebook or Google ad and spend $20. Go to the sub-reddit or discussion board you belong to and put out a post on your idea. Show ten friends your idea. See what results you get.
If you get none, move to the next idea. Fail as quickly as possible. When you find the idea that works, latch onto it. Push it as far as you can.
Online, the future belongs not to those who spend years and years designing the perfect product, but to those who strike idea-oil. They poke here, poke there, and when they get a response, boom! They go all in.