Greedy Goblin

Saturday, June 27, 2009

On healing meters

Sydera quoted Ghostcrawler and did not believe him.

Ghostcrawler said "first Holy priest had much larger healing (total and effective) on the fight than their second Holy priest, so they asked the second priest to go Shadow. They kept wiping. They then swapped them, and made the star Holy priest go Shadow. The second Holy priest’s healing was much lower, but they won on the first try. The second priest just had better timing and cast the right spell at the right moment, even though his total and effective healing was lower overall."

Sydera did not believe him because "I have to say that GC’s Tale of Two Priests seems apocryphal to me. As my guild members pointed out in our website discussion, this is an odd situation that we can’t really imagine in any of our raids. It seems–to us anyway–that there’s not enough information here to judge what really happened. One of my guildies suggested that maybe the “star” Holy Priest was also really good at Shadow, and that seems pretty reasonable to me"

Well, I can explain exactly what happened. Look at the meter below on a 6% try at I choose you Steelbraker:

I suck. Healed 2/3 of our star trial member who already the best of the bests. Most probably the only reason why we did not kill Steelbraker yet is having suckers like me. Recruit one more of such great players, get rid of me and Steelbraker will go down. My only excuse is that the Holy-light-pala and the priests sucks even more!

Except that no one in the raid died. The tanks died, either because of fusion punch or quickly after res. On early tries they died at P1 for God's sake! Let's take a look on the two tanks and see what the suckers did:

See now? No? Let me explain.
  • Rejuvenation ticks 5 times with talents, about 2600 each. Takes 1.2 secs GCD to cast it. So the caster side HPS is 10800 without overheal.
  • Wild growth ticks 7 times, each tick is smaller than previous, they are around 550. Heals 6 people with glyph, takes 1.2 secs GCD to cast. Caster side HPS is 19200! And since it's a smart heal, it finds the lowest health people so it practically doesn't overheal.
  • Nourish heals around 5500, crits around 45%, so with living seed it heals around 7500. Takes 1.5 secs to cast it, so caster side HPS is 5000.
  • Rejuvenation ticks 5 times, about 2600 each, 3 secs between ticks. So the recipient side HPS is 866
  • Wild growth ticks around 550 every secs, so it's not a miracle that the recipient side HPS is 550.
  • Nourish heals all it has into one target and limited by healer cast speed (you can't get them faster than the healer cast them). So recipient side HPS is equal to caster side HPS: 5000.
Two rhetorical questions:
  • Which of the 3 spells has the best chance to save the tank?
  • Which of the 3 spells does the superstar never uses?
Note: please don't comment about "Glyphed HT or Regrowth is better than Nourish", it's not the point.

Once upon a time I wrote a pro-meters article saying "more healing done = better player, period". I still believe I was right, despite it's the exact opposite what I'm suggesting now. Sydera's position is somewhere in between my old and this post and I believe she is right and I really don't blame her for not understanding Ghostcrawler.

How could all three posts be right? Because they are accurate description and good advices in the situation the writer was. When I wrote my first post, I was healing PuGs in Naxx. Sydera healed normal Keepers and normal Yogg attempts. I wrote this post after healing Yogg farm, Hearthbraker kill and (countless) Steelbraker attempts.

The trivial content of Naxx is facerollable. The "best" healer is the fastest faceroller. No decisions need to be made, just spam heal. The meters are perfect measurements of your spamming effectivity. The healer who heals more on the meters is the better healer.

The harder fights need several decisions both in gear choices and spell selection. For example I had 250 more spellpower during Mimiron normal than Heartbraker and 150 less Mp5. Why? Because Mimiron is intensive but have FSR regen periods, Hearbraker is 10 mins long without stop.

When complicated decisions are to be made, "better" or "worse" depend on the effect on the outcome that cannot be measured during one fight. It could be measured by comparing 10 fights where the first healer was present with 10 with the second was not, though such experiment would be quite hard to make.

The McDonalds employee who packs 2x more hamburgers is 2x better than the other employee. The university professor who does 2x more lectures is definitely not 2x better (often not even better at all) than his collegue.

What's the point? Use meters on low-end jobs (spam this or that) and use individual evaluation for high-end jobs. It applies both to WoW and real life.

And yes, it's pretty possible that 4 good spammers could heal Naxx25, while 4 clones of me not, just like most professors create hamburgers slower than McDonalds employees. That's exactly why people geared up grinding 5 mans with 0 raiding experience can be terrible to raiding guilds. The poor guy have to forget the thinking that made him good in 5-man speedruns.


Astmathic said...

And this post is a good example on why you are a good player and are evolving into a really good raider.

Anonymous said...

I think this post demonstrates that you have not learned to evaluate people on how well they perform their healing role.

Your druids, except notably you, were top on healing and very low on tank healing, hence they owned on raid healing. This is as it should be.

Your Disc priest will always look bad on meters because all of the damage they prevent through shields and such is not counted so that is no surprise.

Your CoH priest looks like he was assigned to tank heal, which is fine, but he will then consequently do less healing since he cant use PoH or CoH as effectively. It also looks like you were assigned to tank heal exclusively, which is a waste of a druid, but oh well.

Your pally is top on healing on the dk tank which is perfect, and as it should be.

So you really can't meaningfully evaluate anyone by looking at meters unless you know what role to which they were assigned.

Thunderhorns said...

I think healing meters are deceiving. I generally have lower heal meters always than the MT and OT healers and I'm the raid healer. Yet I don't let people die unless they run out of spell range or something freakish happens.

I play to conserve mana as much as well as heal well. That means I will not spam heal a raid with Chain Heal just because I can. I will LHW the hurt person if no one else is hurt.

It tends to hurt me on the meters, but I'm still a very effective healer. Just like when I'm put on MT heals I know to switch from spamming LHW to mixing in riptides with normal HW as well as Earth Shield.

You use different spells for different jobs and get the same results: victory.

I hate wasting mana spamming AoE heals. It's a cheap way to jump up the meters. I like the way I do it because I heal when needed. I read and learn to expect the spike damage, I heal accordingly conserving as much mana as possible.

Healing like anything else you do is about learning the rhythms of each encounter. When the spike damage will come, how much your tank can take, where people are spread out in a raid, how much AoE damage they will take and what spells are best to deal with it all.

When I see someone dominating the healing meters, I think they are either overhealing or your raid crew brought too many healers. Healing should be enough to push your healers to a work level where there mana and healing ability are hitting the wall and the rest of your raid is able to maximize dps. If you're healers aren't working much and that third healer in ten man or sixth or seventh in 25 man is doing nothing on the heal meters, then maybe it's time to drop one healer down and add another dps.

The only time meters should matter is for dps. And even then you should be concious of dps dropping if someone is handling a task such as add control or interrupts. Same thing for heal meters, raid leaders should be concious of whether a raid is taking much aoe damage to warrant alot of raid healing or what spells the healer is using to keep the raid topped off. Healers that know how to conserve mana, but don't pump the healing meters can be just as good as other healers.

Sephrenia said...

"The tanks died to fusion punch". That says it all for me. Your dispellers suck! No tank should die to a funsion punch, they should get one tick at most.

If people live and the boss dies, the healing team have won. That's all that matters, not who did more. You could look at the meters and see who was most efficient - who healed most with least overhealing?

Have a read of this if you REALLY want to top the meters and never mind the rest... ;)

Anonymous said...

The meters do tell only part of the story. My guess from your post is that you were assigned to tank heal, and the 'star' druid was assigned to raid heal.

Raid healers will always 'win' the meter race.

One of the other really nice abilities of Recount is to look at spell selection. A Disc priest will look 'bad' on the meters, but may actually be your star player. That could very well be the case here...

outdps said...

Hey Gev, I'm fairly sure that that wasn't ghostcrawler's quote. It sounds like something I read over on matticus the other day.

Pullen said...

So we're saying DPS are like McDonald employees? :D

radu said...

Hi. I'm Mooii, from the guild Gevlon is raiding with.

I agree with your overall conclusion:
"What's the point? Use meters on low-end jobs (spam this or that) and use individual evaluation for high-end jobs. It applies both to WoW and real life."

So far, you have pointed out valuable raiding-related information on several occasions, but I do think you need to look at some more in-depth data to draw conclusions. Her's some points I want to make 4 points regarding this issue:


You are looking on a restricted set of data. We reached P3 on 60% of tries (6/10). Out of those 6 tries, 4 where close to a kill. To make any judgments that bring us closer to a kill, one has to analyse these 4 tries where we where close to a kill (tries 5, 7, 8, 9). You can use our WOL reports whenever you want. As you can see the data is identical to your in-game meter, so you can trust it.


We need to look at the critical P3. The rest of the fight is important, but any death in P1 and P2 is a mistake. On the try you analyzed (try 7), you are looking on healing for the entire fight. The situation is different if you look at P3 healing only for that try:


It's hard to say who's heals are the most valuable. I can make a theory and say "For druids, hots are good heals, but to save someone's life, youhave to do a direct heal". So, we can look at direct healing done for that phase, and assume that those diret heals saved lives, while the hots just kept people up.
If my theory was right, I could say that we can rank druid contribution to P3 by how many direct heals they cast, since those saved others from dying.


But, because healing meters measure the allocation of a finite quantity between several competitors, my theory is wrong, and your tank-healing argument is wrong.

If you stop selling glyphs, some other guy will do it, so the market will still be there. If I stop raid healing and start tank healing, I'm sure your hots will heal for more, but you won't be so high on the tanks. If our rejuv-spamming trial stops doing that, you and me will have to do it more.

The only way to form accurate opinion-neutral conclusions on such matters, is, as has been said, to experiment. Since we can't do that, we will have to rely on our officer's and member's opinions of healer behavior.

That being said, I think you are a good healer and I would love it if you could join us on Sunday as well. After ~20 tries of doing a good job on Steelbreaker, you deserve to be there for the kill, and we need our best healers on the job!

Alena said...

I just can't skip that - if that's worldoflogs site that you are using for logs, then it counts disc's absorbs as well. That's to the posters above saying that poor guy is not shown on the meters in all his beaty. :)

Lissanna said...

Where you fall on healing meters depends on what role you are assigned to. Someone raid healing the majority of the time will out-heal tank healers. However, if it's a fight where tanks really need a lot of healing, then whether or not the tank healers are doing their job is not going to be measured by the meter, but by whether or not the tank died (if the tanks death was something that could have been healed through).

I recently went through and talked about Sydera's post & healing meters on my blog.

The total healing done statistic is descriptive, but doesn't actually tell you if 10% healer is meaningfully different from 6% healer. You can see if they are using the types of spells they are supposed to be using in a particular fight, and whether or not they healed who they were supposed to be healing. If your top star druid was assigned to heal a tank, and the tank died... and then they showed up as having not focused on the tank at all, then your top star druid was inflating the meter numbers instead of doing their "job."

Also, druids make great tank healers, with using all their HOTs combined with Nourishes & swiftmends. However, if you just used nourish & didn't put your HOTs on the tank or have the nourish glyph, then you would be doing something wrong.

On Assembly, we pull 2 of the tanks far away from the other bosses, and healers assigned to those 2 tanks automatically fall low on the overall healing done meters, but if their tanks survived, then they did a good job. If their tank died because they ran out of range to try and heal the raid to inflate their meters, then maybe it's time that your guild stopped caring so much about rankings. Sometimes, ranking low on the overall meter says that you filled your role better than being high on the meter.

Healing Meters don't measure skill.

Anonymous said...

""The tanks died to fusion punch". That says it all for me. Your dispellers suck! No tank should die to a funsion punch, they should get one tick at most.

If people live and the boss dies, the healing team have won. That's all that matters, not who did more. You could look at the meters and see who was most efficient - who healed most with least overhealing?"

Well Sphrenia, it sounds to me like you've never bothered to try the fight's Hard Mode on 25 man. Because Once Molgeim and Bruundir are dead, Fusion Punch (The initial direct damage hit, not the dot that needs to be dispelled), hit's really really hard, and Steelbreaker gains a stacking self buff every time a tank dies (which happens thanks to Meltdown, once every 30 seconds). By the time you're on your second tank, Fusion Punch is hitting for 50k+. By the third? 60-70k.

Most kills require 3-4 tanks. Each needs to survive 2 fusion punches.

In order to do so, they need Shield Walls, BoSac's, Guardian Spirit's, Pain Suppressions, etc. etc. But Dispels are the /least/ of the issues.

Brian said...

I think the problem with the healing meter debate is that people like Lissanna correctly identify situations where the healing meters don't tell the whole story, and people like Thunderhorns interpret that to mean that healing meters never tell any part of the story.

The thing is, you can't really "trick" a healing can only heal damage that people are taking. The player at the top of the healing meter isn't padding the meters, he's doing necessary healing to make sure people don't die. If he wasn't doing that much healing, the other healers would have to do more or the raid would die. There's no such thing as unnecessary healing, except overhealing, which generally doesn't count on a healing meter.

Now obviously there are exceptions due to assignments, positioning and the need for AOE vs direct heals, but those are all tools to help you read a healing meter, not reasons to throw it out.

Ranjurm said...

Brian you are wrong to say that healing meters can't be tricked. Let's examine some of the ways it can be done.

Taking additional damage with the purpose of healing it. Frequently a healer can cause themself or others (usually through encounter mechanics) to take additional damage which can be used to raise their numbers. Is a healer who doesn't dispell on hegain for the higher raid damage really doing meaningful healing?

A second means of increasing unnecessary healing done is by sniping heals. Does that dps really need to be topped off with a 4k flash heal now when there's a druid HoT on him and no immediate danger?

A corollary to snipe healing is unneeded topping. That dps with a 1.5k deficit doesn't need a heal from a healer when there's a JoL on the boss.

And there are other ways to do meaningles healing to boost your own numbers on a meter without benefiting the raid and sometimes harming it.

Bell said...

Healing meters are one of the most complicated things to pigeon hole. In the end, the combination of healing meters, healing assignments, healing skills, who-healed-who meters, who was healed by who meters, overheal meters, and dispel meters all have to work in conjunction to understand what is going wrong.

Healing assignments change everything. When there is a wipe and healing comes into question because of tank death, one of the best things to do is to check who is healing your tanks. If someone assigned to the tank is not coming in top or second on healing that tank, they aren't doing their job. If too much of the raid is dying and the AoE healers are on the bottom of the charts, they aren't doing their job (in most cases). If someone who is on a main tank is doing more AoE healing than tank healing, or vice versa, they aren't doing their job. If someone's healing is low but they've been doing 60-70% of all dispelling, then obviously their global cooldowns are being jeopardized and others need to pick up on some of the dispelling or you have to deal with their healing being low. It's also good to remember that perhaps people are dying because, despite the healing, the DPS isn't fast enough to mitigate stacking debuffs or the tanks aren't taunting in rotation or everyone's just standing in the glowing pile of crap.

Even on spam-fests like Naxx, straight healing meters can never tell the full story or say who is a "good healer." I know I've personally found myself on the top of them at times when I had no cause to consider myself having done a good job, even in Naxxramas.

Sydera said...

If you'd read my whole article carefully, you'd realize that I agree with you.

GC's story doesn't give a key piece of information...the assignment. I presume it to be raid healing, which makes the anecdote a little odd. If you actually read my article you'd realize I gave GC the benefit of the doubt despite not having enough info to judge it for myself. The point is, without the actual report, I can't tell you "who's better"--and sometimes not even then.

Brian said...


You're right, it is possible to "trick" the healing meter in little ways. What I really meant was that I don't think tricks like that amount to very much when it comes to judging healing meter results. Despite being a druid healer and healing with a lot of HOTs, I have never been topped on the meters by someone "sniping" my heals. Sure, every once in a while a HOT will get stomped by a fast, direct heal...but that tends to not amount to very much. As for topping off, there are very few situations I've seen where someone takes a small amount of damage and doesn't need to be healed quickly. Of course what you're talking about could happen, but I'm not sure it happens often enough to dismiss healing meters as a whole.

I guess my position on healing meters is formed mostly by experience. I tend to top the meters in raids I heal in, whether I'm assigned to tank or raid heals (mostly because my philosophy on tank healing is that it's not an excuse to stand around with your thumb up your butt if the raid healers need help). And when the OTHER healers in the raid get close to me on the healing meter, it becomes very easy to keep everyone alive. On the other hand, when I'm doing 2x as much healing as every other healer, bad things start to happen.

Obviously healing meters don't tell the whole story, but read correctly, they can tell a big part of it.

Brian said...


GC's story doesn't sound quite right to me for a reason beyond the lack of important details. If the second, lower on the meter, Priest had the same assignment as the first chart-topper...who was healing all the extra damage the second priest WASN'T healing? No matter how good a healer you are, casting the right heal at the right moment CAN'T make up for a significant deficit in overall healing...because eventually all that extra damage is going to start taking people down.

Wooly said...

Healing meters can't be trusted anyway. They give different results and while opinions differ about what meter is best: there are no hard facts. E.g. on recount I end first by a mile on every encounter, and on WWS I fluctuate between first and fourth. I have my own WWS account, so both meters use the same logfile. I tried substracting/adding overhealing on both, but the totals remain different by a mile. So they're nice for a global idea but prove nothing, even if the most healing done would say who is the best healer.

But even though I agree that good performance can't be measured: a meterwhore might do more useless healing/overhealing, as long as his/her effective healing is high you can't deny that there was damage to be healed, and he/she healed it. It could have meant a lot of heals were sniped from others needlessly, but unless he/she is not healing other people on purpose (which makes no sense), this doesn't really matter. He/she is a good healer.

What GC says is the most politically correct answer. It's what he should say, it's what people should keep in the back of their mind. I agree with him for that reason.

But the response Sydera gives does make the most sense. Just the fact a group wins or loses by changing something doesn't mean anything. The average player has the tendency to completely believe, without a shred of doubt, that the latest change to the tactic is THE reason for the succes. Even though the latest chance could even made it worse, but the reason you did win could very well be:

- More luck with the RNG (huge reason imho).
- Random players might have learned a seemingly small new thing from last try.
- Mistakes made by others last run that nobody noticed (not even the one making it) not made this time.
- One more round of practice since last wipe for everyone.

And there are probably more.

Anyway, when I play disc spec I know meters will suck. Things like pain-suppression or inspiration are non-measured mitigation effects. Also shields are not measured/measurable yet. They can all be raid savers.

So, like you say, I also believe the truth is in the middle (isn't it always). I keep this as a general rule: topping the meters can with a reasonable certainty tell that you are a relatively good healer, but not topping them does not say you're a worse healer.

It sounds painfully politically correct also, but I'm convinced it's the truth.

Jeff said...

This is the difference between grasping the subtleties of your class and not. A friend is a discipline healer, if he is alive there is almost no way to kill the five man he is healing, and he can solo heal most of Naxx. He never tops the healing meters.

As a tank I know when to use my abilities and and when not to, I almost always have a taunt ready, an interrupt available, we are not top raiders, we are, in your parlance, socials. We play with a group of our friends who are spread across the country, and we have fun together and clear content. We are slower than a top guild because we only raid 4-6 hours a week, period. But we have very good players who know their classes and rarely have more than one attempt for first boss kill in Naxx (cleared) and three for first boss kill in Ulduar (working Assembly of Iron). We run tens because we don't want to deal with the drama that comes with twenty-fives.

Looking at our specs L337 raiders tell us to go read EJ n00b5. We do, and then modify based on what works best for our group. We build our specs and raid comps based on players (and their play styles) and class synergies.

The basis of our philosophy? Kill bosses with braims(sic). The game is built on math, simply leverage the math as a group and don't stand in the fire and things die.

Who cares what the meters say as long as the boss dies, the only real reflection of the success of a raider.

Thunderhorns said...


You're way off that healing meters can't be tricked.

I know healers that do the following:

1. Druids pre-hot prior to even entering the fight on as many targets as possible even if they aren't taking damage.

2. Shamans spam chain heal even if a person is taking single target damage or only a couple of people are taking damage.

3. Priests spamming CoH every time is it up even if they don't need to.

4. Paladins getting huge overheals on single targets just because their heals do so much extra healing above and beyond what is neeed or healing melee with the Holy Light glyph while other melee are close.

I watch this stuff happen all the time. As well as healers doubling up on heals on the same target because they cast at the same time.

I raid heal. I don't always the top the meters like some say you should. But that's probably because I'm mostly doing 10 mans, and they don't have as much raid damage and there are far fewer players.

But I don't just spam Chain Heal when a raid member gets hurt. I use LHW alot more for fights where individuals are targeted. If a tank is getting hit, I don't pop a heal off every time the guy takes a smidgen of damage.

I see alot of people doing this during raids. A healer sees the tank take a small amount of health and he immediately casts his biggest heal or spam heals the guy. Spam healing is pointless to me. I only do it when it is needed.

Fights have a certain rhythm. Each tank will take damage differently. You have to learn what they can take and manage your heals by it.

Even if you do that, alot of the time in a standard healing crew in a 25 man you may have multiple healers on the same tank and multiple heals may hit the guy at once where only one was needed.

All sorts of things like that happen all the time.

I rarely look at healing meters and I won't say they never give you any information. I didn't say that at all. I said they could be deceiving and they can be.

But you can also see if someone is seriously not pulling their weight or heavily undergeared. Sometimes the meters do give you some red flags.

But alot healers are fairly average and rely on spam healing to be effective. This pumps their meter ratings way up and makes them look alot more studly than they are.

When I'm healing I'm not going to compete on the healing meters. I'm going to manage my mana tightly and heal when necessary. I'm not competing with other healers. That's a waste of our overall heal mana pool. I'm healing as needed and using all the tools I'm given in as efficient a manner as possible. If that means don't top the healing meters, then I could give a rip. It's all about winning.

Since I've been assigned to raid healing, MT healing, and the various healing tasks in Ulduar such as Gauntlet or Arena and succeeded in all of them, I know very well how things work as a healer.

Sydera said...


If you're tank healing, you shouldn't be doing much raid healing, for two reasons.

1. You risk a tank death. Ulduar bosses are all about the big, slow hits, and our tanks have so much avoidance that it only looks like you don't need to watch them. You still do.
2. If your leader has assigned too many people to tank heal, he'll never figure that out unless you stick to your assignment. When you do what you're told, you help out the group leadership and let them refine their choices.

The only way I'd suggest a combo of tank and raid healing is if that is your specific assignment. As a healing lead myself, I never use that one :)

MyName said...


I think what you're saying about not healing groups works great for 25 mans and is no good at all for 10 mans, especially 10 mans with only two healers. Even on Mimiron or Iron Council, if you're tank healing, you still have to help heal yourself and maybe throw a quick heal to help someone else, if nothing else.

That being said, your not likely to need a healing meter in that case as you can pretty much tell who isn't cutting it and why people died.


I have a bit of a problem with the internal logic of your post. On the one hand, you're trying to say most healers are mindless button mashers, but then you're also portraying them as devious enough to go around "tricking the meter" to "pad their score". They can't be both stupid, and devious. If group healing involves alot of button mashing most of the time, and the meter isn't providing useful information, it's more likely that the metric is at fault than that people are tricking it somehow.

Also, I think your analysis of even average healers is lacking as well. Group healers will mash buttons and heal little damage when they're bored or they are not starved for mana. That means they're on trivial content (e.g. Naxx). When they move to fights that are likely to run them out of mana, the button mashing 'magically' stops. On those fights, smart raid leaders will also give out assignments to help prevent overhealing.

It's nice to hear that you are smart about your mana, but most of the time, when people are throwing around alot of heals, it's because they know they don't need to worry about mana on that fight, not because they are slow/stupid/wasteful.

Anonymous said...

healing done does not = good healer imo... different classes/specs have different roles, for example holy paladins will not have as much healing as say a resto druid or a holy priest, because their "niche" is and always has been tank healing (even though a smart paladin can heal the raid with beacon on the tank, the focus still needs to be the MT) so as long as that kind of healer manages to keep the tank up, and helps out in raid healing, he's doing his job. But the numbers don't show that.

yuripup said...

And even your spamfest-Naxx healing who ever is highest is best is wrong.

I am on the US East Coast and play on a West Coast server. My latency is about 150 ms higher than the average west coast connection (mine is at 200-250).

That means I am seeing the damage around 0.15 seconds after some of the other priests--and their facerolling (is there anyone not elite you don't show contempt for) is faster than mine.

So again, the heal meters don't show what you think they do.

And again "finding the data" takes a lot of opinionating and making judgments as to what is good and bad data.