Last time I observed that neutrality doesn't work. Even if Initial Neutrality does work by leveling the field, it does so by making the learning task too hard in general, which isn't good. (I've seen projects that were deliberately made unnecessarily hard for the sake of neutrality.) Outcome Neutrality is good in theory, but people vary, so what works for some doesn't work for others.
This leads us to the Eterna Komencanto or Eternal Beginner. Though most noticeable among Esperantists (simply because there are more of them), such people can be found in all auxlangs large enough to be subject to the laws of probability. And even a little reflection shows that the phenomenon is even more common among natlang learners.
Think about it: we've all seen (even perhaps been) the one who just doesn't get some language. I think I first personally witnessed it in high-school French, though there were indicative incidents in Spanish class as well. The intensity of the condition ranges from lack of confidence for active use to total inability to learn, though passive ability is generally good enough to understand at least the gist of a text.
Someone will no doubt object that anyone can learn any language, if it's properly taught. That's technically true. But there's a difference between technicality and practicality: the only reason some people know even their native language is from years of uninterrupted immersion as children--an extreme measure impractical for most adults, especially if the language has relatively few speakers. In theory, any auxlang could be forced on everyone. In practice, people will only willingly learn an auxlang whose difficulty does not exceed their interest, so an auxlang must be incredibly easy or incredibly interesting. And since interest requires either practicality (and thus a large user base that will not appear out of nowhere overnight) or a personal appeal that will be present for some people but not for others, it's a shaky rationale anyway.
Given all this, we may ask where the global user base for any given auxlang will come from. Most projects I've seen simply assume that users will appear; the projects are often ill-suited to attracting users strongly enough to invest the time and effort to master the auxlang. "Hey, that's clever" isn't enough.
And consider that many people, though strongly motivated to learn some natlang, barely achieve intelligibility. English has plenty of Eternaj Komencantoj, probably more than there are Esperantists of any kind.
The conclusion? Some people will always be left out of any given auxlang, but it won't be quite the same group for all auxlangs taken as a whole. This will normally be considered unacceptable, but I'll argue next that it's completely acceptable once we drop the Onelang idea of a single auxlang for everyone.