On Class Distribution in Patch 5.2

Screen Shot 2013-02-19 at 12.21.24 AM

I publish any graph with a great deal of trepidation; there are caveats and collection methodology and a lot of footnotes which go into serious data analysis which seem to always get lost in a single graph presented without context. But the context is often vital to avoid misinterpretation.

Patch 5.2 is coming soon, and I’ve returned to the data collection I started while writing The Decline and Fall of Warlocks in Cataclysm to see how things look for Warlocks. One important thing I learned last year was that most Warcraft population sites are focused on current, not historical, data. This requires ‘snapshotting’ data at critical points to allow for trend analysis. This is the third snapshot I’ve taken so far. Furthermore, each site has quirks and variations which make it impossible to reconcile them exactly. We can use them to talk about general trends – as long as we do so skeptically.

Instead of continuing to update the file used in Decline and Fall, I’ve created a new spreadsheet for this discussion. You can follow along in the Google Doc if you like.

I’m calling this snapshot the 5.2 patch data even though we’re not officially in 5.2 yet. There won’t be any massive shifts in population in the next week or three.


Screen Shot 2013-02-19 at 12.22.04 AM

I’ve assembled the data from 5 time periods in a summary format to see very general trends with class movement in Mists of Pandaria, with Wrath (patch 3.3.5) and Cataclysm (4.3.3) data thrown in for comparison. This has some advantages for general population trending but is also problematic in other ways, which I’ll discuss later.

This chart, and the accompanying graph at the top of this post, are the relative population component of endgame characters for a given time period. This means that the first 3 columns represent level 85s, while the last two are level 90 characters only.

When I first made this type of chart last year it compared apples to apples – namely, end of expansion population figures. Players had had time to level alts and have multiple toons at endgame, so the data represented mains and alts alike. The Mists snapshots are critically different in that they are at the beginning and middle of an expansion, when leveling time is limited and content is fresh and demanding.

Understanding this difference in data type is critical to avoid making hasty judgements based on these numbers. There’s a pretty big disconnect between the first two and last two columns because of this end-of-expac effect.

  • Wrath and Cataclysm numbers are end of expansion numbers and represent mains and alts alike.
  • Mists Pre-release represents level 85 characters before Monks or Pandaren were introduced, in patch 5.0.4. This is the final snapshot of Cataclysm, and could be considered roughly equivalent to the previous two.
  • Mists Patch 5.1.0a gives us our first level 90 data. This specific data isn’t 85-90 data – it’s level 90 data. This snapshot shows us who leveled to 90 during the first few weeks, and is probably the best data point we’ll have for divining which classes players considered to be their mains, even if they switch later on.
  • Mists Patch 5.2 is another level 90 data point, this time with additional alts and slower levelers joining the endgame. Having 3-5 level 90 characters is not uncommon in this snapshot, so now we’re seeing who else people play.

There’s an additional complication in that Monks were introduced between 5.0.4 and 5.1.0a, shifting the average popularity from 10% to 9.091%. This ~1% drop can be adjusted for with indexing popularity values, but it’s not really worth it at this stage in the expansion. It’s not worth it for two reasons: because the 5.1.0a figures represent the actual popularity of main choices, so indexing isn’t appropriate yet, and because the effort to level a Monk to 90 is still substantially higher than leveling any other class from 85 to 90. Over time this will even out (and even skew towards the Monk class as their leveling bonuses come into play), but for now the imbalance should stand.

A quick look at the data shows that since Cataclysm:

  • Warlocks got a little more popular.
  • Warriors got a lot more popular.
  • Mages and Rogues were common alts at the end of Cataclysm, but not mains, or they main switched.
  • Paladins seem to be mains but not first-tier alts, as evidenced by the relative slide in standings since 5.1. Druids (and possibly Warriors) seem to have the same issue.
  • Hunters, Warlocks, and Monks seem to be gaining popularity as alts heading into 5.2.

The drop in Rogue popularity seems to be that we’re seeing the core of the class emerge – the die hard Rogue mains who will stick with it no matter what. The Legendary daggers offered to Rogues at the end of Cataclysm artificially inflated their numbers, but we could see other players level those Rogues to 90 by the end of the expac. I feel comfortable saying this because the 85 and lower data doesn’t show a drop at all.


Screen Shot 2013-02-19 at 12.29.14 AM

The above data is from realmpop. I like realmpop a lot because it allows me to go through and drill down into the data, so that if I wanted to find out how many female goblin death knights are still stuck in the starter zone I could do it. The drawback is that the results are graphical and split by region, so I have to manually copy the values and add up populations between the US and EU. The sample size is large enough that I feel comfortable using relative values like popularity, but I wouldn’t want to use them for absolute values like server population.

If you unhide the columns on the second tab of the spreadsheet you can see the raw data from each snapshot.

The reason I think Rogues aren’t in any new state of crisis is because of the data above. When you look at the class across all levels, they’re pretty solid (and don’t show any decline.) But as soon as you get past level 85, the numbers fall off precipitously. People haven’t wanted to level them – yet. Perhaps they leveled one so their guild could get the legendary daggers in Dragon Soul. Perhaps they saw how they were performing in PvP and switched (more on that later.) But they’re there – just not at the endgame.

There’s a different set of problems there, of course. Why is there this drop off? Why do people not want to level Rogues to 90 but do want to level Paladins or Shamans or Warriors instead? There are problems here, but they’re not as simple as the problems affecting Warlocks in Cata.

I’ll leave that up to the Rogue bloggers to discuss, but I expect Blade Flurry has something to do with it. My own Rogue has been stuck at level 67 forever.


Screen Shot 2013-02-18 at 11.51.51 PM

One of the other data sites I’ve come to rely upon has been World of Wargraphs; like realmpop, it pulls data directly from Blizzard’s API (not through in-game addons, like Warcraft Realms), but it presents the data in very different and interesting ways. The PvE/PvP breakdowns, in particular, are very helpful in determining what specs are over- or under-represented in high end play.

The next four tabs on the spreadsheet are dedicated to snapshotting the heroic raiding and 2200+ PvP class and spec breakdown. Some of the lists are rather long, so I’ll provide direct links here:

Screen Shot 2013-02-18 at 11.54.41 PM

A few things to note.

  • Guardian Druids are not present in the data as a separate spec.
  • Arms Warriors are an amazing 11.4% of all 2200+ characters surveyed. I think we found our missing Rogues.
  • There are some specs which are struggling in both environments. Unholy DKs, Demonology Warlocks, Holy Priests, Marksmanship Hunters, even Fire Mages could use a look.
  • Hunters, in general, seem to be having problems at endgame. It could be a number of reasons –  perhaps it’s that they’re easy to get to 90, but hard to master in raids and PvP alike. Perhaps they’re too complicated to play well at 90. Or perhaps it’s that they are favored alts for dailies? I honestly don’t know.
  • A few specs are doing well in both environments. Holy Paladins, Shadow Priests.

Really, the biggest story from this data is how overwhelmingly popular Warriors have become for ranked level 90 PvP, and how scarce Rogues have become in that same activity. I think these trends are absolutely related.


Overall, the changes to the Warlock class in Mists seem to have had a positive effect on relative popularity. Players are rolling Warlocks and leveling them to endgame. This is a massive improvement!

Affliction and Destruction are reasonably represented in PvE and PvP. Demonology seems to be less common in high level play, but one of the current Arena world champions won playing Demo, so I don’t know what to really say to that yet. Perhaps it’s just that it’s really tricky to master? Don’t know.

Patch 5.2 presents something we haven’t seen in a while – Warlock-only quests. There’s a lot of interest around the green fire quests which will no doubt prompt people to try leveling one to 90 to give them a try. This kind of attention can be good if the class fundamentals are sound, which I think they are again. But it’s going to skew numbers in the future.

We need to collectively remember that when looking at the class later on.



Filed under Cynwise's Warcraft Manual

18 responses to “On Class Distribution in Patch 5.2

  1. Fire mages dipped due to the nerfs, I’d definitely say. Especially as you go higher up in progression levels. They were so good at the start of Pandaria and then the changes made it hard not to just go Arcane in a broken fashion.

    • Parker

      Yeah, I raided as a fire mage in cata and had great numbers, full reg DS clear, but now in Mists it feels like I’m just sitting and watching my Living Bomb tick down, I don’t get Pyroblast! procs as often as I did, it’s become quite dull. I sometimes play frost when I feel like a change but more often I just play a different toon.

  2. One item of note is that it seems that Mists is gaining a reputation for being very alt-unfriendly. If that truly is the case –and I’ve not gotten my new main to L85 to find out personally– then all of those Rogue alts might simply remain lower level alts.

    I’ve found that as I’ve leveled my Rogue via BGs, there certainly doesn’t seem to be a lack of them in BG areas. I suspect that since Rogues are viewed as primarily a PvP/BG class, their appeal doesn’t necessarily extend into normal PvE content. It’s been a while since I’ve seen a raid group that goes out of their way to look for a Rogue, for example.

    • I agree – this is my ‘cry-baby’ response, but personally, I leveled so many alts, and now feel that Blizzard still has a traditional response to mains versus alts, instead of just one player at the keyboard. And after reading this, I’m very concerned about paladins–that seems like a significant drop in character choice. Maybe that’ll be your next thesis, “The Rise and Fall of Paladins!” Is there a way to garner data for each player, and a slash/played sort of chart, to see what characters are played the most? I think I’ll go do that now…I’m sure I wrote down my /played data somewhere….:)

      Oh, and the female goblin Death Knights – you may want to add in the “pet collector” factor, too, and CRZ – it was damn near impossible to get a scourged whelping without a realm or starting a DK on a Brazilian server.

  3. Informative as always, Cyn.

    Your data about locks — and in particular, demonology — confirms my suspicions and experiences from the last couple of months. Demo is hard. I haven’t taken the time to really analyze it by the breakdown of how many things there are to remember like you did in DAFoWIC, but it goes something like: 1. Curse of Elements 2. go demon form 3. Doom 4. back from demon form 5. Corruption 6. Hand of Gul’dan 7. spam Shadowbolt 8. Molten Core? spam Soul Fire 9. Demonic Fury +900? go demonform, spam Touch of Chaos. Oh, and throw in cool-downs like Dark Soul, imp swarm, the Doomguard and your trinkets.

    And that’s just the opening 15-25 sec of a fight in PVE! Juggling dots, and managing 2 resources (mana and demonic fury) and a pet is a lot of work! The only saving graces are that the spec is actually fun despite itself, and has lots of “cool factor” (shapeshifting, best lock pet), and as a guy can only run LFR, my dps seem competitive. I guess if it’s fun, Bliz can call it a success (in light of DAFoW), but I won’t be surprised to see some streamlining down the line.

    • I really have a hard time believing that Demonology is weak in PvP after watching Snutz play it at the recent championship. Difficult? You better believe it. But really potent if you master it, and that’s a good thing.

      I should give Demo a good solid try one of these weeks. It usually takes me about 3 weeks to really switch specs, so I have to be prepared to suck for at least that long before giving up on it.

      • Without a doubt! I just wanted to reiterate that I don’t think Demo is weak in pvp, cause I don’t think it is. When you talked about specs that are “struggling”, i thought you meant in representation, not performance (granted the two are probably closely related). Your illustration about Snutz is probably more telling than you think. Let’s say he’s (she? idk) the best in the world. So how long should it take just an average player to master Demo? 6 months? A year? That’s what I’m expecting. Don’t know if my alt-aholism can handle that, haha, but then I don’t really deserve to be a top ‘lock.

        Another point about class complexity and how Blizzard “teaches” us to play. Compare the Demo “what has changed?” tab in-game to http://www.icy-veins.com/demonology-warlock-wow-pve-dps-rotation-cooldowns-abilities.

      • FWIW, once a Demo Lock goes Demon Form, that’s when I’m circling around to backstab it. What I’ve found is that some Locks use Demon Form as a kind of bridge between other attacks, so (from an L40s Rogue’s standpoint) I look at Demon Form as a temporary weak link in the chain.

        Now, that might change once I get to higher levels, but it seems that a lot of Demo Locks in mid-level PvP are more focused on getting out Wild Imps than anything else. I can understand the temptation, because a pack of Wild Imps can create a big juggernaut of a convoy, but focusing on Wild Imps leaves a lot of abilities unused.

  4. Kal

    I think that part of the problem with Hunters is the button bloat, and certain changes to the class. While we now have great self heals/damage mitigation and Kill Command has a greater range, some of the new abilities have fundamentally changed the way our rotations work. The level 60 talents tend to drown us in focus, and the level 75 and 90 talents add too many buttons to press. Powershot and Glaive Toss were bugged, adding further frustration (fixed in 5.2). I haven’t even leveled my own Hunter to 90 yet, mostly because I’m waiting for all these changes to settle before switching back from my Pally and Rogue.

    My Rogue is my second most geared character now, so I obviously like playing it. I have my complaints, but I think the class is moving in the right direction. I wouldn’t call it my main, but it was one of my first three to 90, while my Hunter “main” is still 87.

    My chief complaint about the class is how slow it is in PvE. I’m quite used to button mashing, having mostly played a Hunter since Wrath. Rogues do not button mash, apparently. I think this is chiefly due to Rogues passive energy regen, but also because finishing moves cost energy. While a Hunter is always actively building or spending focus, a Rogue waits for energy to build combo points, then waits for energy to spend them, then waits for energy to build them back up. Honestly, I think the best thing I can do to increase DPS is just staying on the boss butt, maximizing my auto-attacks, poisons, Main Gauche, ect. Such a passive nature to DPS really doesn’t square with me, and may eventually cause me to focus only on PvP.

  5. This is amazing data Cyn, but I’m having a little bit of heartburn over the conclusions. I can’t help but feel that there’s a lack of context with these numbers (though that’s absolutely no fault of yours), because what you’re capturing is characters at various stages of leveling, but not activity levels.

    If we are going to look at the popularity of a class, I would imagine that time investment might be a better metric (or at least a contrasting one), on the premise that classes that players enjoy are ones that they spend more time on. The point at which you level that character doesn’t seem all that tied to your preference, only the perception you had of the class at that point. (Although, I suppose there’s an argument for sustained interest if you get them all the way to 90). My alt druid, for example, was leveled to 90 within the first week of the expansion, but has remained largely ignored for most of the tier, in favor of my third alt, a priest. In this vein, abandonment numbers might be even more telling. But unfortunately, this data really isn’t accessible outside of Blizzard’s offices.

    And, I’m not sure how you judge popularity in an environment where I think you could make a reasonable argument that favorable imbalance will generally trump appealing playstyle. So, for example, the large prevalence of disc priests (over holy priests) in heroic raids isn’t because we all love the healing style more, it’s because discs are incredibly powerful at the moment.

    In that regard, I think it would be an interesting angle to look at spec popularity versus DPS/HPS performance. Because I’d hypothesize that your high-performing classes would also be your most “popular”, and therefore the outliers of that comparison (high performance-low pop or low performance-high pop) would point you towards classes where other factors were at play.

    Sorry for the long comment, but heartburn or not, I think this is the best style of post–one that inspires curiosity! Many kudos.

    • I would do very very bad things to get my hands on that kind of data. Being able to construct queries like “show me the characters that people spent more than 1 hour on last month … okay, now break down those toons into percent time played for the account… okay, now look at time spent in raids versus time in daily zones.” … I would do bad things for that data. Bad, bad things. The data we have access to is so frustratingly opaque.

      My own playstyle means that I’m not counted in many of these graphs at all, despite happily rediscovering my main and playing 95% of my time on her now. All you’ll see is another level 85 Warlock. We make assumptions about leveling and playtime and they’re not necessarily the right ones.

      I think the PvE/PvP numbers are highly indicative of when a spec is overpowered for the job, while general level 90 numbers are more reflective of people’s enjoyment of the class. It’s good to call that out as an assumption. There’s a dichotomy here that we have to keep in mind, though – some people get attached to characters and won’t change no matter what. I think we’re seeing that with Rogues now – only the diehards are left, and they don’t care if the FOTM swings another way. They’re going to play their Rogue until they shut off the server. So we have to give it some leniency both ways.

      Man. /played data per month, by zone. That would be sick.

      • In my professional life when we need data we go out and get it; why not ask your readers for this? Bad things optional.

        • Sample size would be too small it the primary problem. Both of the sites I use have databases with 30-35 million characters in them; even if I muster a few thousand responses there’d still be a lot of response bias. D:

          If the API could be changed to return this data, though…

        • Actually – I thought about this some more and think you have a point. I shouldn’t be so quick to dismiss it. Thanks!

          • Of course! I’m quick with ideas, but the execution is the daunting task; let me know if I can help. You already have legions of loyal readers, some might even say minions, to assist you in your own quest for data! 🙂

  6. Thanks again for your analysis Cyn. I was considering looking into the numbers again largely due to rogue discussions. I was looking at your spreadsheets a while ago in light of some discussions around it – one of the interesting statistics is that rogues actually have the second highest numbers in the level 10-20 bracket (after hunters) – they’re just very popular twinks, and/or people hate levelling them.

    But I would agree that rogue’s are sitting in a different place to warlocks were. They’re numbers were definitely inflated by the legendary factor, and they’ll have been even more inflated because the legendaries were introduced in the last patch of the expansion. The shift to no legendaries, but more importantly from high to low haste actually means that the game play is terribly different (Feral dps I think suffers from it even more however!) I understand it because I was in that boat – however my rogue is my most geared character now – I love the play style and don’t want to give it up. I was also probably the least affected by this because I largely play Subtlety and it is probably the fastest played of the three specs.

    Rogues though had a very negative general consensus from the MoP beta going forward which would have put a lot of people off – mostly due to their slowness and survivability/damage in PvP. It’ll be interesting to see if there is a change in rogue numbers with the 5.2 changes.

    The last statistic I’d like to see in terms of rogues is also the numbers in comparison to the other similar classes – though that use energy – feral druids and now windwalker monks. I wonder if a log of the popularity around them is tied to the resource type (a lot of people point towards on target combo points – but I think it is more likely that it is energy that is the culprit). Hunters in theory are in the same boat – but the reality is they have focus building tools and it isn’t exactly the same. Energy as a resource allows you to do a lot at once but has the cost of then having to wait for it to rebuild. I really like it because it works well with latency, and it gives you a chance to watch and see what is happening at the same time. But it does leave big gaps in the rotation and this makes it feel slow. While it can be rhythmical that rhythm isn’t enforced by cooldowns. Rogues versus monks and ferals however are locked into that playstyle in all specs of the class.

    In terms of access to the data – have you tried throwing gin Ghostcrawler’s way? 🙂

  7. I have noticed an increase in warlocks in my guild as well. I’m really pleased to see more people becoming interested in playing the best class. I’ve sadly stopped raiding with my demo lock. But this is because of raid balance, we’ve already got 2 others in our 10 man group.

  8. I’m one of the rogue diehards I am the only person in my guild raiding on a rogue main, and its been my main since vanilla.

    Its hard to compare raiding rogues to paladins, warriors, shaman, and druids when it comes to raiding. All of those other classes have tank or healing specs. That will surely push their numbers up a bit in comparison to rogues since they can fill multiple roles.

    Mages, hunters, and warlocks are the other pure dps specs. Comparing rogues to those classes eliminates the healer and tank factor. The main difference – rogues are the only melee pure dps class.

    In the raiding game right now, with the exception of a few fights, ranged classes are more enjoyable to play than my rogue. When you have to run here, move there, don’t stand in this, and go stand in that, rogue mobility can be frustrating. I’ve had times where I wanted to quit raiding on my rogue (Will of the Emperor comes to mind). If I thought that, as dedicated as I am to my rogue, then I am sure that others have thought the same.

    Just my thoughts.