As this expansion comes to its denouement, with Heroic Garrosh vanquished and a long wait until Warlords of Draenor is released, I’d like to take some time to reflect. This has been the first expansion in which I have completed all of the content “pre-nerf”, as well as my first expansion with a serious progression guild – and the most I’ve ever raided in 25-player mode – so the whole experience has been exciting and new. I also have never thought quite so much about healing, especially in regards to the “big picture”, as I have in MoP! And since I like to write, I’m afraid I’ve ended up with a healer manifesto, of sorts, which I will share with you over the next few weeks.
Those of you who come here seeking math and data, well, there will be some of that, but most of what I’ll focus on in this series of posts is my opinion, and a lot of that is predicated on how I feel about the healing game. The posts will focus primarily on the 25-player raid-healing experience, since Warlords of Draenor is skewing towards larger raid groups with its flexible content and 20-player Mythic modes, but I will spend some time on 10-player issues as well. I’ll tell some stories, take a look back at the trials and tribulations of each healing class, talk a little about what worked and a lot about what didn’t, then take a look forward into the future – or past? – of the next expansion.
This is going to be a pretty huge undertaking! I’d like to consider a number of topics from the perspective of the entire healing milieu, laying down the groundwork and base assumptions behind my opinions, and spend a post on each healing spec really drilling in to their performance throughout MoP and my hopes for WoD. So this may sprawl out to a huge number of posts, but hey, I have a lot of thoughts😛
Nothing’s Quite the Same as the Healing Game
Those of you who have followed me for a while now are probably not surprised to hear that I have not been exactly enamoured of the healing game in MoP. While encounter design has been fantastic, and the focus on making personal responsibility a more integral part of success in raiding has definitely improved the end game, healing itself has felt rather shallow and, at times, unfun. This was most apparent for me in my 25-player raid – perhaps because the healing game in 25s is always inherently skewed toward AoE healing as the best course of action, making half our toolkits irrelevant – but I did feel it to a lesser extent in my 10-player raids.
Actually, at first it was great. Early tier 14, when we were undergeared and pushing Heroic progression, felt amazing. I didn’t have the mana to cast Healing Rain on CD, so I had to be more careful with its use. I couldn’t just spam Chain Heal between Healing Rain casts – and back then, I had to think harder about when to cast it, because it wasn’t worth it on fewer than three targets, and targeting was a real issue.
The other healers in my raid had similar constraints on their mana, and that made each of us feel like our decisions and skill made more of a difference. Knowing when to use that throughput cooldown on Heroic Garalon, or how to ration our healing so that we had enough mana available for the final 20% of Heroic Gara’jal, or when to switch to single-target healing to save lives on Heroic Stone Guard – these things aren’t irrelevant now, but they’re a lot less important. And as a person who appreciates the challenge of making smart decisions and working as a team with my healing colleagues far more than the challenge of topping the meters, I live for this kind of gameplay. It is what drives me to learn the nuts and bolts of every fight, to use that knowledge to make more informed and effective healing choices, and to eventually write ridiculously thorough encounter guides😉
As the expansion progressed, though, this exciting feeling faded. I felt more like my decisions didn’t matter. I could cast Healing Rain on cooldown now, so there weren’t really decisions to make other than where to place the reticule. Smart healing was so spammable and abundant that I no longer had to think about who to heal – the question was only when; with the legendary meta gem, the answer to this question was “as often as possible”. Cooldown choreography trivialised more mechanics, leaving less and less for throughput healers to actually heal. And absorption effects seemed out of control, easily dominating meters and serving as the most readily visible scapegoat for all of our woes. But are these perception problems, or actual problems? In either case, how can we fix the problems and make healing more engaging for the next expansion?
These introductory posts are going to cover each of these points, but the first thing I want to tackle is an overview of healer representation. This is an important metric as it can reflect issues of both aesthetic and pragmatic preference, and because I think that ideally all specs should feel they are able to contribute to success in every part of the end game. I’d love to see healers judged not by the colour of their raidframe, but by the content of their character sheet. Or something.😛 And watching the healer population fluctuate can itself tell an interesting story about the real influence of the flavour-of-the-month mindset and an ever-changing balance landscape.
So what does the data say? To answer this question, I went to Raidbots and created graphs showing the number of players of each healing spec that show up in World of Logs each week by selecting “All Parses” as the Data Set and “Samples” as the Measure. I then adjusted the Timespan in the URL until I was able to isolate (mostly) just the MoP data.
The red lines on these graphs indicate major patches, and patches 5.2 and 5.4 brought new raid tiers – Throne of Thunder and Siege of Orgrimmar respectively. The yellow lines indicate the dates for which I collected specific data to calculate percent representation of classes, which you’ll find in my ‘Normal versus Heroic‘ section below. For most of this discussion, I’ll be focusing on Heroic raids – Heroic encounters tend to be highly demanding on healers, whilst strict enrage timers or DPS checks encourage raids to drop to as few healers as possible, and often add extra mechanics that can affect healer performance. As a result, healer representation is possibly more fluid in Heroic raids, and we may see the effects of buffs, nerfs, and encounter mechanics more clearly in this data.
I should also state that this is not the most scientific data set ever collected. A class’s representation in raids is not strictly about their performance – perhaps that class is just particularly popular (or unpopular) for other reasons. Perhaps community perception puts pressure on players to stick with one particular spec over another. In choosing Heroics, where performance can make a difference, I hope to have eliminated some – but of course, not all – of these “aesthetic” biases, since Heroic raid leaders are likely to pick their healing composition based on pragmatic reasons. But these are interesting data points as well. We don’t necessarily have to understand why the data is the way it is in order to get value out of it. We can still look at fluctuations in each spec’s representation throughout the tiers, and at relative changes, without having to standardise for overall class population amongst the entirety of WoW players.
I want to be clear: The Raidbots graphs that illustrate my points in this post are not depicting the class’s throughput. They are demonstrating healer representation. And the point to everything that follows is not to complain about healer balance – in many respects, healer balance in this expansion has been the best it’s ever been. I don’t have an agenda here – I just find this stuff interesting🙂
25-Player Heroic Healer Representation
This graph shows the progression of healing representation in 25-player Heroic raids over the course of the expansion.
You’ll notice that, back in tier 14, there was a pretty huge gulf between the three most popular specs – Shaman, Disc Priests, and Paladins – and the rest. Well, at first Disc Priests were outweighed by Monks – who were quite frankly OP at launch – but Disc Priests took off dramatically after being overbuffed with Patch 5.1, and Monks didn’t keep up the dramatic rise that Shaman and Paladins did. Monks and Druids stagnated at fairly low representation, and Holy Priests were extremely rare.
The reason that Paladins and Disc Priests were so dominant in tier 14 is the same reason they remained dominant for the rest of the expansion – absorption effects are amazingly powerful and are especially useful for Heroic encounters. By comparison, Holy Priests were even more underrepresented than usual in t14 because low Spirit and a lack of other intrinsic mana regeneration capabilities meant this spec suffered the most from mana problems. While Holy throughput has always been excellent, they’ve always been a poorly represented spec in Heroic raids since high-level, world-first raiding guilds prefer the effective health and raid CD benefits of Discipline Priests, and the rest of the community tends to follow suit.
I’d suggest that the only reason Holy Priests show up as much as they do in this t14 data is because of Heroic Tsulong. Holy Priests were in fact the most popular healer for 25-player Heroic Tsulong, due to the strength of Guardian Spirit as a raid-wide throughput CD for Tsulong healing, and Divine Hymn in Night Phase was pretty great too.
Monk representation for Heroic Tsulong was also anomalously high (compared to the overall data), due to the strength of Life Cocoon and the fact that Monks, with their 1-second GCD and instant spells while channeling Soothing Mists, could get off more spells during The Light of Day than any other healer, making them a great choice for Tsulong heal-bombing.
The reason Shaman were so popular in t14 is a slightly different and more nuanced story. Early in an expansion, Shaman have three huge advantages – their Mastery, which compensates greatly for low-level gear; their numerous throughput CDs that have additional value when undergearing or underhealing a raid; and their inherent mana regeneration capabilities (Water Shield, Resurgence, and if used, glyphed Totemic Recall) allowed them to perform at closer to maximum capacity at earlier gearing stages. These advantages are likely to be present again in the first raid tier of Warlords, and with Shaman currently being highly represented, I think we’ll see Shaman come out quite strong at the start of the expansion.
Druids were extremely weak at launch, particularly in 25-player raids. Their newest tool, Wild Mushrooms, just wasn’t terribly easy to use, and their traditional palette of healing spells just couldn’t compete with the new tools given out to other classes for AoE healing. Rejuvenation spam and Wild Growth didn’t really cut it compared to extra Healing Rain buffs, the Priest Level 90 talents, Light’s Hammer, or for that matter Uplift. This problem plagued Druids until tier 16 and the introduction of the Glyph of Efflorescence, granting Druids access to a persistent AoE effect (and we’ll talk more about this later, too).
You can see that in tier 15 there was no longer a huge gulf between specs. Paladins suddenly overtook Discipline Priests, perhaps due to encounter mechanics favouring Spirit Shell a little less and Paladins becoming more capable of keeping up a tight Eternal Flame rotation thanks to having enough mana regen in their gear to cast expensive Holy Power generators. Shaman declined as – with a few notable exceptions, namely Megaera and Iron Qon – stacking just wasn’t much of an option in this tier.
Meanwhile, Druids and Monks stayed at relatively the same population as they did in tier 14, and Holy Priests jumped up a fair bit as Spirit regen and the Legendary Meta Gem made their mana situation a little more palatable. I expected to see more Holy Priests in Heroic Tortos, much like we saw for Heroic Tsulong, but frankly Disc Priests were just fine for that encounter too, so there wasn’t as huge of a swing as I’d anticipated. Still more Holy Priests than Disc Priests, but not by the significant margin we saw earlier:
In Tier 16, we return to having a large gulf again, with all the other healers leaving Monks and Holy Priests behind. Druids have leapt up for the aforementioned reason, and Shaman are dominating because there are many fights with tight enrage mechanics for which dropping healers is an attractive progression option, and oh yeah, because we got overbuffed to compensate for the demoralising experience of t15.
Drilling down to the special-healing-mechanic fight, Heroic Malkorok, we can see that Holy Priests now don’t even outnumber Disc Priests. In fairness, this encounter was designed to be kinder to Disc Priests in allowing absorption effects to stack on top of (and “outside” of, therefore protecting) Ancient Barriers, so it’s no surprise that those Disc Priests who were swapping to Holy for this fight at the start of the expansion now don’t seem to be bothering.
I know this representation data looks pretty bleak for Monks, but let’s remember that Monks are a new class, and new classes typically have slow growth when they’re first introduced. It’s also important to note that the mechanics of the class may not necessarily be appealing to Heroic raids – or to Heroic raiders. Class mechanics changes intended to make Monks choose between “Fistweaving” and “Mistweaving” styles, plus the nerf to Revival and the inherent weakness of the Renewing Mist mechanic and its interaction with Uplift for burst AoE heals, have left the class a little lacklustre for 25-player raids.
10-Player Heroic Healer Representation
This graph shows the progression of healing representation in 10-player Heroic raids over the course of the expansion.
Here we can see that Paladins and Discipline Priests are the undisputed royalty of 10-player Heroic raiding. It’s not surprising – the same things that make absorption effects desirable in 25-player raiding make them indispensable in 10-player raiding. 10-player raids are more likely to have a small dedicated healer core with an additional player who offspec heals for 3-heal encounters, and in this scenario the raid is best served by having two absorption healers with the most powerful damage reduction raid CDs as the dedicated healers, with a throughput healer filling the ‘swing’ role and contributing powerful healing CDs even when they are DPSing, like Healing Tide Totem, Ancestral Guidance, or Tranquility.
Discipline Priests do exceptionally well in 10-player raids because they can maintain high healing throughput whilst also contributing a fair amount of damage. If a Disc Priest can approximate a quarter of the damage throughput of a DPS, well, this is a much bigger deal when there are only 5-6 other DPS in your raid than when there are 16-17 other DPS in your raid. They also bring two raid cooldowns (if you count Spirit Shell, and you should, since in 10-player raids it can easily rival healer throughput cooldowns like Tranquility). And for Paladins, well, it’s easier to keep Eternal Flame rolling on more targets, which until 5.4 kept Illuminated Healing refreshed on the entire raid with ease, and Paladin utility like Devotion Aura, Hand of Protection, and Hand of Sacrifice is also extremely valuable in the smaller format where fewer raid CDs are typically available.
For most of this expansion, everyone else has taken a backseat. However, in tier 16 – as with the 25-player data above – class representation for Disc Priests, Holy Paladins, Shaman, and Druids is really even. In this regard, I would have to view the buffs granted to Druids and Shaman in patch 5.4 as a success – allowing them to even up the score with Disc Priests and Paladins is quite the achievement considering the gulf between these specs’ representations in previous tiers. Unfortunately, Monks are left behind again, probably for the same reasons as I listed above.
Holy Priests are all but nonexistent in 10-player mode, even in those fights with mechanics that might favour Holy Priests, like H Tortos or H Malkorok, whereas in 25-player mode we see a smallish but competitive Holy Priest population for these encounters. Holy Priests were well-represented in the 10H Tsulong encounter, as the second-most popular healer brought along, but outside of that fight the spec doesn’t show up much. The pressure for a Priest to be Discipline in 10-player raids is really, really high. There are obviously outliers – as any Priest has probably heard, the first 10H Garrosh kill was solo-healed by a Holy Priest – but the average 10-player Heroic raid contains approximately zero Holy Priests.
By choosing Heroic encounters I have necessarily biased this data towards smaller sample sizes and performance-based class selections. While I won’t obsessively analyse Normal raid healer representation, I thought I’d just present here some comparison data, as a series of pie charts showing comparative healer representation throughout the expansion. To generate this data, I chose one date from each tier – which corresponded to strong 25H activity – and recorded the number of logs for each healer spec on that date, then calculated the percentage representation of each spec as a whole. The results…
… probably won’t surprise you. There is not a vast difference here. Normal-mode raiders are more likely to play Holy Priests than Heroic raiders are, because the value of Discipline is not exaggerated as much in Normal modes. Not to mention, while there certainly are a lot of progression- and min/max-minded Normal-mode raiders, there are also quite a few Normal-mode raids with a more casual, play-what-you-want kind of attitude, where players won’t feel as pressured to move away from their preferred spec based strictly on the community’s perception of its performance.
I find it a little interesting that Monks are slightly more represented in Heroic modes than in Normal modes, and wonder if this isn’t related to the power of Monks at launch. Perhaps Heroic raids are more likely to request a player swap specs or even classes in order to pursue a better composition?
These effects are pretty much even in both raid sizes – so I think we can expect to see it remain similar in Warlords.
Warlords of Diversity?
Based on this data, healer diversity actually looks pretty good, with the exception of Monks and Holy Priests. Remember, diversity doesn’t necessarily mean that healer balance is fine – this data alone can’t say much to that – but it’s heartening to see other specs catching up to Disc Priests in 10-player raids. However, there is a significant limitation here – some Heroic modes are easier to kill than others, and therefore logs of these encounters contribute the majority of the data in this sample.
If those easier Heroics are also particularly suited to using more healers, because they have looser enrage timers or are simply first killed by undergeared groups that require more healers to keep up with the damage, then it’s possible that this will bias our data. For example, if most 10-player Heroic raids have a two-healer core of, let’s say, a Priest and a Paladin, but tend to have a third class off-spec heal for the few three-heal fights, and those three-heal fights are also the easiest and most frequently killed Heroic bosses, then we will see a picture of healer representation that appears to be more balanced than it is.
To combat this, since there is greater variation in total healer numbers in 25-player raids than in 10-player raids, I have also looked at some harder, and more traditionally underhealed, fights from this expansion to see if the representation is vastly different to the overall picture I’ve painted above with the tier-wide data. Since the stated plan for Mythic raiding is to balance it for 5 healers, I suspect that Mythic raiding diversity will look a fair bit like underhealed-25-player-Heroic-encounter healer diversity, so this could be pretty instructive as to what we’ll see in raid groups in Warlords.
Unfortunately, the best examples of underhealed-25-player-Heroic encounters I can come up with simply aren’t available on the Raidbots website, because World of Logs has not yet implemented Heroic detection for them – Garrosh Hellscream and Siegecrafter Blackfuse. Most Blackfuse kills were with 4 to 5 healers, and most Garrosh kills include 4 healers, with some guilds dropping down to as few as 3! There were a few commonly underhealed fights in previous tiers though – Heroic Gara’jal the Spiritbinder was 4-healed, and Heroic Lei Shen was often 4- or 5-healed, and we have nice Raidbots graphs for these.
In both of these cases, the class representation is fairly close to their respective tier’s class representation as shown in the original 25-player representation graph above. Notably, Monks got a boost on the Lei Shen encounter compared to their overall representation. The mechanics of the fight heavily favoured having one Monk healer to stick around in melee during phase 3, able to heal the melee and tanks during the Violent Gale Winds/Thunderstruck confluence – and also because Monks’ powerful personal survivability tools were invaluable for handling frequent Static Shock events. By comparison, Shaman with their area-limited heals and (at the time) lack of powerful smart-healing abilities were less represented on this fight than usual. But in general – compared to their overall representation in these raids as a whole – no one of these classes was wildly overrepresented or underrepresented in these underhealed fights.
From what I can gather about the two underhealed fights at the end of tier 16, though, things aren’t quite as egalitarian. I checked the number and spec identity of healers in the “Execution” ranks in Warcraft Logs‘ listings, which are sorted first by number of deaths and within those categories by amount of damage taken. This method was chosen primarily because it’s clear these kills were well executed, so clearly the reduced number of healers is the correct number of healers. And the data is not biased inherently towards absorption healers, since Warcraft Logs counts absorbed damage as damage taken. (You could argue there is still inherent absorption bias since players are less likely to die when affected by absorption effects; a random sample would have been better, but I couldn’t find a way to quickly generate a truly random sample.)
For 25-player Heroic Siegecrafter Blackfuse, I queried 49 individual kills. Of those 49 kills, 31 were 4-healed, and 18 were 5-healed, with 214 unique healers. For 25-player Heroic Garrosh, of the 45 kills queried, 43 were 4-healed (1 was 3-healed and 1 was 5-healed), with 180 unique healers.
|Healer Representation in Underhealed Heroic Encounters|
This data looks a fair bit less even than the 25-player overall representation data shown before, though the trend of “pretty good for everyone except Monks and Holy Priests” remains.
Discipline Priests are more popular in these two fights than they are in the rest of the data I’ve collected for tier 16 – and are the most popular class to be doubled-up on in these kills’ healer comps. Not to mention, there was not a single kill in my Heroic Garrosh data that did not have a Disc Priest, and only one Heroic Siegecrafter kill without a Disc Priest.
These two encounters were some of the hardest in the tier, with a lot of unavoidable damage that Discipline Priests are really good at preventing, clear mechanics for Power Word: Barrier to mitigate, and tight enrage timers for Atonement to assist with, so it really isn’t surprising. But this data may – may – be evidence that Discipline Priests are too powerful for end-of-expansion progression. If this is true, there are a lot of possible factors behind it, and we’ll look at them more in the mana regeneration and absorption posts of this series, and again in the Priest-specific discussion.
Since Discipline Priests are so strong for these encounters, it is no surprise that Holy Priests suffer the greatest drop-off in representation.
I’ll note that Shaman are more represented in Garrosh than in Siegecrafter since the predominant 25-player strategy involves stacking in Desecrated Weapon during phase 1 and stacking on Garrosh through Whirling Corruption and Empowered Whirling Corruption in phases 2 and 3. The fight is an excellent example of where the Shaman stacked-healing-and-lots-of-CDs niche can really shine. Monks are less valuable since they bring only one raid CD and it is quite weak. (And many 25-player strategies use a Brewmaster Monk to kite Minions of Y’Shaarj for the final half of phase 2 and the entirety of phase 3, so it is possible at least that some Mistweavers are respeccing Brewmaster to serve this function.) Cooldown diversity seems to be appreciated by Garrosh-killing raids as well, as the most common healing composition was Druid/Paladin/Disc Priest/Shaman.
By comparison, the mechanics and damage pattern of the Siegecrafter encounter favour Paladins over Shaman, where the damage is less frequent, often occurs in one-off blasts (Overload) or to single targets (Superheated), and is more readily mitigated by absorption effects and Devotion Aura than by Healing Rain and Healing Stream Totem. That said, the predominant strategy for this encounter does involve quite a bit of stacking, so Shaman are still well-represented, just less so than in Garrosh kills.
This data also kind of suggests that when a raid needs to increase its number of healers, it skews towards Holy Priests and perhaps Shaman to do so. Since both of these specs are reactive, throughput-type healers with powerful raid cooldowns, this is a pretty sensible option. Fights where you need more than the ‘standard’ number of healers – for example, Heroic Grand Empress Shek’zeer or Heroic Thok the Bloodthirsty – typically involve sustained heavy damage for which these classes – and their raid CDs – are ideally suited.
I’m well aware that this information hasn’t broken any new ground. The community seems to know intuitively that Discipline Priests and Holy Paladins are very powerful and well-represented (although remember, we cannot draw conclusions about the power, or overpoweredness, of a spec solely from representation data). But I don’t like to work without data backing me up, and it’s important to confirm our preconceptions before talking about whether there is a problem and, if so, how to fix it.
I certainly didn’t expect to see a perfectly even split between all six specs, so the fact that Monks and Holy Priests don’t each constitute 17% of the raid-healing population is not necessarily indicative of a problem. It’s more that the healer composition used for underhealed content is so skewed towards specific specs that indicates a problem. Something is making those specs more valuable to the raid than the others, and that is a cause for concern.
If my hypothesis holds – that Mythic raiding healer composition is most likely to resemble underhealed 25H healer composition – then I think we’ll see, moving forward into WoD, the persistence of a tendency towards Druid/Paladin/Disc Priest/Shaman comps. If Warlords doesn’t change whatever underlying factors are responsible for low Monk and Holy Priest representation, it’s quite possible that we’ll see a majority of raids without any Monk or Holy Priest healers at all! And yet paradoxically, as raids move to larger sizes, the prospects for Holy Priests do improve.
So can the developers make Monks and Holy Priests attractive enough to show up as more than single-digit representation percentages in Mythic raiding? To achieve this goal for Priests, this necessarily means the relative power imbalance – perceived or actual – between the Disc and Holy specs must be addressed.
For Monks, this is a difficult task to accomplish, because the class is new and has a unique playstyle that may not appeal to the community at large. Great players may pick up the class and raid competently on it without external assistance, but there are a lot of Very Good players who may be held back by the newness of the class and the relative dearth of community guidance. (This is getting a lot better now, but at the start of the expansion there was precious little information.) Mistweavers are slated for a significant overhaul in Warlords of Draenor, and this should address the mechanistic issues plaguing them, but I’ll get into that more in my Monk-specific post.
Another way to approach the problem would be to build Mythic encounters where these specs’ unique tools are very valuable. Since the developers will be designing for a single raid size, instead of two disparate raid sizes with conflicting demands, it is my hope that they’ll be free to put in mechanics that utilise and encourage greater class diversity – leading to better outcomes for everyone.