Someone once said:
First, percentages are a far greater measure of disparity than numbers. Numbers without context are meaningless. The percentage is meaningful. You don’t use the percentage because it makes your argument stupid at best.
Which is incorrect. Neither percentage values nor specific numbers mean anything without context. Neither are a meaningful base unless taken in context.
In Warlords of Draenor, we had 11 classes.
IF, in the midst of WoD, all classes held the exact same amount of max level characters, that would mean that each class made up for ~9.09% of the total character base in the game, at that time.
In Legion, they added another class, the Demon Hunter. Giving us a total of 12 classes.
This means that, even if the total amount of max level characters was still 11 million, the actual percentage value would be lower for each class despite there being no change in the specific number of characters at max level.
100% divided by 11 classes = ~9.09% for each class.
100% divided by 12 classes = ~8.33% for each class.
That equals a drop of ~0.76% for each class, even if the actual number of max level characters remains unchanged(which in itself, is unlikely).
Also worth noting is that, since that 12th class did not exist prior, it didn’t exactly drop in numbers, it only gained in numbers.
Some captions:
In WoD(june 2016) = Hunters made for 12.7% of all max level characters.
In Legion(june 2017) = Hunters made for 9.9% of all max level characters.
In BfA(june 2019) = Hunters made for 10.1% of all max level characters.
In BfA(dec 2019) = Hunters made for 8.6% of all max level characters.
Caption comparisons - all classes:
Based on 11 million total max level characters.
Below, you have the percentage value as well as the (number of max level characters)
Warlords(june 2016)
Hunter - 12.7% (1 397 000)
Druid - 11% (1 210 000)
Paladin - 10.7% (1 177 000)
Warrior - 9.8% (1 078 000)
Death Knight - 9.1% (1 001 000)
Priest - 8.4% (924 000)
Mage - 8.8% (968 000)
Rogue - 7.4% (814 000)
Warlock - 8% (880 000)
Shaman - 8.1% (891 000)
Monk - 5.9% (649 000)
Total: 99.99%
Legion(june 2017)
Hunter - 9.9% (1 089 000)
Druid - 10.9% (1 199 000)
Paladin - 10.3% (1 133 000)
Warrior - 8.8% (968 000)
Death Knight - 7.4% (814 000)
Priest - 7.1% (781 000)
Mage - 8.3% (913 000)
Rogue - 5.8% (638 000)
Warlock - 6.6% (726 000)
Shaman - 7.8% (858 000)
Monk - 5.4% (594 000)
Demon Hunter - 11.6% (1 276 000)
Total: 99.99%
Comparison difference from Warlords - Legion
Hunter: -2.8% (-308 000)
Druid: -0.1% (-11 000)
Paladin: -0.4% (-44 000)
Warrior: -1.0% (-110 000)
Death Knight: -1.7% (-187 000)
Priest: -1.3% (-143 000)
Mage: -0.5% (-55 000)
Rogue: -1.6% (-176 000)
Warlock: -1,4% (-154 000)
Shaman: -0.3% (-33 000)
Monk: -0.5% (-55 000)
Demon Hunter: +11.6% (+1 276 000)
Total, WoD-classes: -11.6% (-1 276 000)
What does that last part mean?
Well, WoD classes in Legion saw a total of 1 276 000 fewer max level characters compared to the same period in Warlords of Draenor.
Demon Hunters at the time of the caption in Legion, held a total of 1 276 000 max level characters.
Does this mean that the reason the other classes saw fewer characters at max level, that this was the result of how all those had moved to the Demon Hunter class?
No. Not at all.
Max level characters does not equal the amount of active players/subs in the game. Some had probably rerolled to play only Demon Hunter. But we have no indication of exactly how many there were that did this. Some probably played multiple max level characters, incl. Demon Hunters.
In short, you can’t point to Global Statistics when discussing how many players(subs) a particular class has lost. You can however deduce that, as an example, fewer players were interested in playing a Hunter at max level in Legion, compared to the same period in WoD.
We can see that several classes, incl. Warrior, DK, Priest, Rogue, and Warlock were played less at max level as well. Although, none of which come even close to that of the Hunter class.
This tells us that the Hunter class was an outlier for this period.
Was there only 1 reason for why this was the case? Nope.
But considering the Hunter class was the one to be changed the most, is it wrong to take the removal of RSV in favor of the addition of MSV as a big factor for this result? Again, nope.
Note that the 11 million value, is only used to add a perspective to the percentage here. We have no actual clue of the exact number at the time in Legion.