I get Shang's point and I agree with the gist of it.
It's Dark Shaman that seems out of the blue and nonsensical to me as any type of wheat from chaff separation.
For most* guilds Sha of Pride is the first stumbling block because it's mechanically dense but has a low DPS check, so while not hard, it's going to take a few more tries, I think Sha of Pride is a good indicator of a guild that's a bit more dedicated at this point.
Once you get Sha of Pride there's nothing more numerically or mechanically challenging until Malkorok. That's not to say guilds don't have more trouble on Nazgrim or any other fight as they might have a weak/bad comp, or a bad strat, or whatever. But usually Sha of Pride is Hurdle #1, Malkorok is Hurdle #2, Thok is Hurdle #3 and each boss after that is really its own Hurdle.
Okay so what's the point of looking at this anyway? To see how many mages are in MoP? Well, one set of data isn't going to be useful here. I think it's far more prudent and helpful for analysis to look at a few categories and interpret reasonably from then on:
1) How many mages are level 90?
2) How many mages have killed Garrosh on LFR?
3) How many mages have killed Garrosh on Flex?
4) How many mages have killed Immerseus on Normal?
5) How many mages have killed Malkorok on Normal? (this data point should be a significant drop)
6) How many mages have killed Garrosh on Normal?
7) How many mages have killed Immerseus on Heroic?
8) How many mages have killed Sha of Pride on Heroic? (this data point should be a significant drop)
9) How many mages have killed Malkorok on Heroic? (again, significant)
10) How many mages have killed Thok on Heroic? (again, significant)
11) How many mages have killed Garrosh on Heroic? (probably a terrifyingly low number)
Then plot all those on a chart vs. other classes and you'll get a much better picture of where mages stand in raid representation, how long they survive in progression, and if there's really any trend at all vs. other classes.
No one data point will tell the story here.
*- most does not imply all, it implies most-- don't need a snowflake guild to chime in here thinking they break the statistical curve