The big problem is that no one looks at the potential for these games objectively. Look at historical precedents and guess how many copies big games can actually sell. I'll do some key ones.
Game name Best-case scenario / Fairly realistic expectations / Worst-case scenario
Wii Play 30+ million / 17.5 million / 10 million
Zelda: Twilight Princess 6 million / 5 million / 4 million
Super Mario Galaxy 12 million / 9.5 million / 7 million
Wii Health Pack 10 million / 3 million / less than 1 million
Metroid Prime 3 3.25 million / 2 million / 1.5 million
Super Smash Bros. Brawl 8 million / 7 million / 5 million
Animal Crossing Wii 8 million / 4 million / 2 million
Mario Kart Wii 9 million / 7 million / 5 million
Nintendogs Wii 6 million / 2.5 million / 1 million
F-Zero Wii 1.5 million / 750K / 250K
Pikmin Wii 2 million / 1.25 million / 750K
Only if you put guesses out there at the actual sales potential of games can we discuss which games are important or not. Anything which does under 3 million units can't really be called a "killer app," as far as I'm concerned.
"[Our former customers] are unable to find software which they WANT to play."
"The way to solve this problem lies in how to communicate what kind of games [they CAN play]."
Satoru Iwata, Nintendo President. Only slightly paraphrased.
