On your points, I agree mostly but in this case people didn't want better quality. They liked the taste of the lower quality product.
It may use ingredients that score better on certain metrics (low sodium, etc.) but these don't mean"better quality" in the sense that these consumers are interested in (the taste that they've come to like.)
Basically, it is blindly tasting three samples, and trying to tell which one of the three is different. (The other two are the same.) So, if you are testing a new supplier, you can give the subjects two samples from the old supplier and one from the new supplier, and see if they can pick out the odd one with any kind of statistical significance.
The key factor is that it is not attempting to measure "which tastes better", but rather, "which tastes noticeably different".
If people were satisfied with the product, changing the ingredients is definitely going to add risk. I worked for a food manufacturer, once upon a time, and we did extensive triangle testing even when sourcing ingredients from a new vendor.
People want quality. That doesn't mean their taste is the same as yours...