the numbers on dual specs

April 30, 2009

These are the correct figures on dual spec adoption. The mistake in the original post was mine (those left outer joins will get you every time) and not a fault of the armoury. The sample size is small, so the sampling error may be large, but the general pattern  is clear enough from these numbers.

Sample size: 30K level 80s who have been played since 3.1. Choices below 2% are not shown.

Class
Specs
Popularity
Death Knight Frost and Unholy 10%
Death Knight Blood and Frost 8%
Death Knight Blood and Unholy 7%
Death Knight Two of Frost 3%
Death Knight Two of Unholy 3%
Death Knight One Unholy 28%
Death Knight One Blood 21%
Death Knight One Frost 17%
Druid Balance and Restoration 23%
Druid Feral Combat and Restoration 18%
Druid Balance and Feral Combat 5%
Druid Two of Feral Combat 7%
Druid One Restoration 16%
Druid One Feral Combat 14%
Druid One Balance 13%
Hunter Beast Mastery and Survival 10%
Hunter Marksmanship and Survival 7%
Hunter Beast Mastery and Marksmanship 3%
Hunter Two of Survival 5%
Hunter One Survival 42%
Hunter One Beast Mastery 21%
Hunter One Marksmanship 12%
Mage Arcane and Fire 11%
Mage Fire and Frost 11%
Mage Arcane and Frost 8%
Mage Two of Fire 3%
Mage One Fire 29%
Mage One Arcane 22%
Mage One Frost 15%
Paladin Holy and Retribution 19%
Paladin Protection and Retribution 16%
Paladin Holy and Protection 11%
Paladin One Retribution 21%
Paladin One Holy 16%
Paladin One Protection 13%
Priest Holy and Shadow 26%
Priest Discipline and Holy 12%
Priest Discipline and Shadow 11%
Priest One Holy 22%
Priest One Shadow 16%
Priest One Discipline 8%
Rogue Assassination and Combat 15%
Rogue Combat and Subtlety 5%
Rogue Two of Assassination 9%
Rogue One Combat 32%
Rogue One Assassination 28%
Rogue One Subtlety 7%
Shaman Elemental and Restoration 28%
Shaman Enhancement and Restoration 18%
Shaman Elemental and Enhancement 4%
Shaman Two of Restoration 3%
Shaman One Elemental 15%
Shaman One Enhancement 15%
Shaman One Restoration 15%
Warlock Affliction and Demonology 12%
Warlock Affliction and Destruction 8%
Warlock Demonology and Destruction 8%
Warlock Two of Demonology 3%
Warlock One Demonology 26%
Warlock One Affliction 22%
Warlock One Destruction 16%
Warrior Arms and Protection 18%
Warrior Fury and Protection 16%
Warrior Arms and Fury 6%
Warrior Two of Arms 3%
Warrior One Protection 25%
Warrior One Fury 15%
Warrior One Arms 14%

Oddly enough, my conclusions are exactly the same as last time despite the different numbers:

  • Take-up of dual specs is much lower than I was expecting. I published figures a while back which showed that 30% of level 80s did all three of PvE raiding, BG PvPing and Arena PvPing and the vast majority of level 80s did at least some mixture of PvE and PvP. I would have thought that anybody who did PvE and PvP would want two specs – which basically means everybody.
    Perhaps people are waiting for the talent-resetting patches to stop. Perhaps the need to lay down the one large (in the Tony Soprano sense) is holding people back. Or maybe it is just too early to see where this feature is going to take us.
  • The most enthusiastic adopters of dual specs are players who want a DPS-and-support combination, where “support” means tanking or healing. Nothing surprising there. Take-up amongst the pure DPS classes is lower on the whole. So only 25% of hunters have a dual spec but 50% of druids do.
  • One limitation of my data is that I can’t yet sort the actual builds into raid and PvP builds. If I could do that, the picture we see here might be a bit different. An interesting question is what pure DPS classes are doing with dual specs. There seems to be two alternatives: a) having the off-spec as a spec-in-test and b) having a PvE and PvP DPS spec. So when we see that a warlock has two demonology builds, that could just mean that they are testing the second build or that they are trying to run as demonology in two different areas of the game. I’m still working on a solution to identifying builds by their intended role so I can get an answer to this question.

10 Responses to “the numbers on dual specs”

  1. Ray Says:

    How do you know that half of these lvl80’s aren’t alts?
    That could explain the low % of dual-spec.

    You say they’ve been played since 3.1 but that only means someone logged on for a minute right?

  2. Flyv Says:

    It’d be interesting to work out what fraction of the non-dual-spec folks have even played since patch 3.1 dropped. I don’t think they publish the last login dates. But you could go by whether they’ve got any new 3.1 achievements (Noblegarden, Tournament, Ulduar.. maybe PvP?). Alternately if you have a pre-3.1 snapshot of the data, any change in the “Total damage done” statistic is a decent indicator of whether they’ve played much.

    Also, you can see if they even trained dual specs. It’s a specific achievement.

  3. zardoz Says:

    Ray:
    Yes, good point. A certain percentage of these characters will turn out to be alts and I should have mentioned that. I’ve got some data which suggests that the played level 80 population is generally pretty active (with a lot of alts still left back in the 70-79 bracket) but no doubt that is a factor. Overall, I’m still sticking to my early-days-yet theory but…

  4. zardoz Says:

    Flyv:
    The last login date is in the XML, and I use it to filter out characters who have not been played recently. Sorting the mains from the alts is the interesting problem since some “active” toons will turn out to be one-trick-ponies like BG PvPers, or just very well equipped crafters and bank mules. That point about using the total damage done is certainly on the right track. Unfortunately I don’t maintain historic data, but maybe I’m going to have to to address this problem.

    BTW, apologies for not yet replying to your email – certain RL issues have cropped up over the past week. I’ll get a reply off this weekend.

  5. ribby47 Says:

    I’m just spitballing here, but I’d venture a guess on your Double-Demo Warlocks as a Raid and Solo build. For soloing, Demo + Siphon Life (21/50/0) is awesome, and a lot of people spec into it. The problem is, going that deeply into Affliction means that you can’t do much for Raid DPS once the initial curses are up as all of the Shadowbolt goodness is more or less overlooked. Thus, build #2 is probably something like 0/51/20 to allow the ‘Lock to pop off some kind of nuke in between renewing DoTs. Just a hunch, given that most PvP ‘Locks seem to favor affliction.
    Speaking from a personal standpoint, I’m kind of considering doing something like this in the short term since a fresh 80 can really fudge some Spellpower out of their gear with Demo.

  6. zardoz Says:

    I’m running some queries to see if there are patterns in the sets of builds now. We may see that there are some popular (x/y/z + a/b/c) combinations. But I think there’s no substitute for finding a way to split the sample into groups by their preferred activity. It really would be a step forward to group dual specs into, eg, (raid & pvp), (BG pvp and arena pvp), (raid and solo questing) etc etc.

  7. Rylia Says:

    The thing that several dps classes in my guild are doing with dual spec is having one build with replenishment, and one for pure dps, switching between on a given raid night to make sure we have the right amount of replenishment. It’d be interesting to see how prevalent that was.

  8. com Says:

    You might be able to infer the purpose of specs by the gear the character is wearing when that spec is active, at least regarding PVP/Raid specs in the same tree.

  9. Gemini Says:

    I’m a SV/BM hunter. SV is my raid spec, and BM is my semi-tanking spec. I focus more on having my pet hold aggro as best as possible and have as much stamina and armor, plus I have the glyph of mending to keep him up more. I use this spec to solo vanilla raid content.

  10. zardoz Says:

    @com

    Yes indeed. Dividing up the sample into raid, bg-pvp and arena-pvp based on gear is high on my to-do list.


Leave a comment