Data testing is the one thing.
Realising lately that data accuracy is the 20% that will give us the 80% improvement in reliability.
Functions and APIs that write correct data to the database resolve most of your major bugs.
It's easy to fix a front-end bug that has correct underlying data.
It's hard to fix a bug that has logic flaws in writing data(backend). Especially if you already have bad data in production(Live).
From this.
I think have been doing these things wrong on Bubble.
Relying too much on front-end functionality
I handle most calculations and things on the front end, which limits my ability to unit test these workflows.
If I were to put this on the backend and set up an endpoint, I could test it programmatically.
This is all pretty raw, and I am still fleshing these ideas out.