I give my EE friends a weird look when I see they bought stuff from Adafruit. I understand if you are a casual trying to do embedded for the first time, but these people know what pins are. They know what MCUs are. They know what power and ground are.. Not sure how they come to the conclusion to spend 10x more.
Here's the deal for me, cost-wise. I can go on Sparkfun, Adafruit, Digikey, Arrow, etc and get these known compatible pieces of essentially Lego that already have firmware support/libraries ready to go. Digikey (who stocks both Adafruit and Sparkfun parts) will have them on my doorstep tomorrow morning. Using the existing libraries, I'll have the a skeleton of a system up and running by tomorrow afternoon. I'll have the prototype up and running in 3 days.
Then, if everything seems to look like it's going to work well and I want to pursue the project further, I'll fire up Eagle or CircuitStudio and lay out a board. Upload it to Macrofab and wait 3 weeks for my fully assembled boards to arrive. And then port the firmware from my prototype over to the assembled boards. Or rewrite the firmware, depending on the quality of the specific sensor libraries.
It's all about bootstrapping the process. When I get a board from Adafruit or Sparkfun, I've got a pretty good belief that it's going to be high quality and fit well within their existing ecosystem of parts and libraries. Building a prototype out of those parts helps de-risk the project and accelerate the project by cutting out most of the up-front dicking around to get it working.
Plus, the vast majority of these boards also have schematics and part numbers provided. When I go to design a real board, I'll use those as references.