One reason
might be that he posted it on a couple of sites that have
SynthV threads. If I remember correctly, he
specifically told people they might be interested in listening
because of the spoken part, and not because they were particularly interested in hearing the song.
So after a minute, they'd heard enough to get an idea what the spoken part was, and they left.
That's just a
theory though. A
music theory.

That's actually a pretty good theory.
I completely forgot to take these posts into account.
I think that solves this particular case. Thanks David

And I like the general approach (keeping the listener's attention, where you place the climax and/or chorus and why) and everything else that's been brought up here in this thread, because everyone is a little different, has different approaches, and I learn something from each of you.
So, even if I haven't replied to every single answer, a big thank you to all of you!