I'm afraid that I'll have to disagree on the basis of the life cycle length and something I call "revision relevance".Obviously the GBA was popular when you compare it to a phone with no games (I like the N-Gage's design but it had next to no support from developpers). But compared to the likes of the Game Boy or the DS, it did kinda poorly.
The Game Boy line spawned four revisions (Game Boy, Game Boy Pocket, Game Boy Light, Game Boy Colour), the last being the Game Boy Colour and its life cycle spanned between 1989 and until 2001 (technically discontinued in 2003, more on that later) - that's 12 years, and on top of that, users had good reasons to update to the Game Boy Colour - exclusive games and a colour display not only allowing to play new games in colour but also the oldies by attaching palettes.
The GBA on the other hand only had three revisions - the GBA, the GBA SP and the somewhat gimmicky GBA Micro which had no lasting appeal to the broad audience with a life cycle between 2001 and 2004 (technically discontinued in 2008, more on that later) - that's only 3 years. On top of that, users had absolutely no reason to upgrade other than their fancy - no new features were introduced with the revisions other than the backlight.
It's also worth to mention that once upgrading from the GB/GBC to the GBA or from the GBA to the NDS, the need for the previous system was automatically gone due to backwards compatibility so I'm going to equate the release of the next line as the "death" of the old one. Whatever was sold from that point onwards were merely stragglers.
As you can see, there are good reasons why the Game Boy line sold 118 million units and the Game Boy Advance sold 81, but once you put the time factor into place, you realize that the Game Boy sold 9,8 million units yearly while the Game Boy Advance sold 27 million units yearly - it was selling immensly better.