Basically, you check yes or no for everyone, and the winner is the one with the most "yes" votes. Less expressive than IRV, but also simpler to understand and tabulate.
Belatedly: thanks for sharing. It does seem an easier sell, because the counting process is so easy to understand. It's not as favorable to third parties as IRV, because "[X] Minor [X] Major" obviously doesn't carry as much weight as "[1] Minor [2] Major", but at least it still kills the spoiler effect. While IRV seems more "pure" in an information theory sense, Approval is elegant in its simplicity.
That's mostly my views as well, yeah. There are a couple (narrow) weird cases with IRV, too... but IRV is unquestionably a tremendous improvement over FPTP.