How the grade is built
METHODOLOGY
Every Umbra grade is a composite of five or six weighted categories. Weights change by role: a tank and a DPS player aren't graded on the same things. Certain specs (Augmentation Evoker) also get spec-specific adjustments where their contribution doesn't show up on their own damage bar. The whole engine is documented here so there's no black box.
Role weights
WHAT COUNTS FOR WHOM
Category scores are 0-100 and feed the composite at these weights. Rows that don't apply to a role (e.g. healing throughput for a tank) are simply absent. The remaining weights redistribute.
| Category | DPS | Healer | Tank |
|---|---|---|---|
| Damage Output | 30% | 20% | 25% |
| Utility | 20% | 20% | 15% |
| Survivability | 25% | 20% | 25% |
| Cooldown Usage | 15% | 10% | 20% |
| Casts Per Minute | 10% | 10% | 15% |
| Healing Throughput | — | 20% | — |
Categories
WHAT EACH ONE MEASURES
Damage Output
dps • healer • tank
For every run, we pull Warcraft Logs' percentile rank for your spec computed against other parses at the same key level. Your +6 is graded against other +6s, your +12 against other +12s. The run page also surfaces a second number showing where the same DPS would rank against the entire spec pool. That second number is for context only and does not affect your grade. Augmentation Evokers are a special case: their personal DPS ceiling is lower by design because they spend GCDs buffing teammates, so we also credit a portion of teammate damage done during their Ebon Might and Prescience windows back to their score.
Improve
Review top-parse logs for your spec at your key level to see their opener, cooldown alignment, and target priority. Most sub-50 percentiles are fixable by tightening rotation, not by pushing gear.
Healing Throughput
healer
WCL percentile for HPS against same-spec healers in this M+ zone. A high score here reflects actually healing (throughput), separate from your utility and damage.
Improve
Low throughput on high keys usually means either the tank is taking unnecessary damage (addressable via CC/kicks) or you're not using your group heals on CD.
Utility
dps • healer • tank
We count every interrupt, dispel, and CC cast across your tracked runs. Classes without a dispel aren't penalized for not having one. Critical-priority interrupts (boss heals, mass CC) count 1.5× when we have the data. Healers are weighted toward dispels; DPS and tanks toward interrupts and CC.
Improve
The single biggest single-player improvement in M+ is kicking the right casts. Install a kick-announcer, learn which casts matter per dungeon, and rotate with your team.
Survivability
dps • healer • tank
Three signals: how often you died, how much of your damage taken came from ability IDs we know are avoidable, and how much healing you consumed relative to the group. Tanks aren't penalized for healing received (they're supposed to tank). Deaths caused by avoidable abilities cost extra. The death penalty softens at higher key levels (above +12) because even top players take unavoidable deaths in high keys. Avoidable damage and healing burden thresholds both scale with key level so a +25 isn't compared against +5 standards.
Improve
Audit your Details damage-taken log after a wipe. If one ability hit you for >15% of your total intake, look up the mechanic on Wowhead and you'll usually find the out.
Cooldown Usage
dps • healer • tank
For each of your spec's major CDs we compute expected uses based on fight duration, then compare against actual casts. Talent-optional CDs are excluded: we don't penalize you for not taking a specific talent. Missing only one CD drops your score; hoarding defensives or offensives is a common sub-grade cause.
Improve
Bind your majors prominently. If you're sitting a 2-minute CD through a 30-minute key, that's 10+ minutes of wasted throughput.
Casts Per Minute
dps • healer • tank
Total casts divided by fight duration, scored against role and spec benchmarks. We use spec-specific thresholds: a 25 CPM Marksmanship Hunter is doing great, a 25 CPM Fury Warrior is struggling. Low CPM usually means rotation gaps, bad positioning, or AFK on a mechanic.
Improve
Check your Details timeline for long gaps between casts. Often these map to movement, mechanics, or a dead target (precast before the pull ends).
Bracket-fair damage scoring
JUDGED AGAINST THE KEYS YOU ACTUALLY RAN
Per-run damage scoring compares you to other parses at the same key level. Your +6 is graded against other +6s, your +12 against other +12s. Without bracketing, a clean +6 lands at 5/100 against a global pool dominated by +12+ pushers, burying low-key performers even when they played well.
The run page also surfaces a second number: where the same DPS would rank against the entire spec pool. That one is context only and does not affect your grade. The two together tell you both how you played the keys you ran and where you stand against the broader meta, without conflating the two.
Key level weighting
HIGHER KEYS COUNT MORE
Inside each category, per-run scores are weight-averaged by keystone level. A +15 carries about 3× the weight of a +5. That's because pushing higher keys is the actual measure of skill. The scoring engine rewards it directly instead of asking us to infer it after the fact.
Support spec adjustments
FAIR TO AUGMENTATION
Augmentation Evokers don't top personal DPS meters because their job is amplifying teammates' damage via Ebon Might and Prescience. We measure that uplift directly (summing teammate damage done during their buff windows) and blend it into their damage score. A top-uplift Aug gets the credit a pure-DPS scorer would miss.
Per-dungeon grades
WHERE YOUR GRADE COMES FROM
The composite hides which dungeons you're actually weak in. Every player page shows a separate grade per active-season dungeon, computed by running the exact same scoring engine over just that dungeon's runs. A tile with a C+ next to your overall A- tells you exactly where the drag is.
Pull-by-pull + rotation
THE WHOLE DUNGEON, 30 SECONDS
Click any run and you get two tabs. Pull-by-Pull shows what happened each pack: kicks, avoidable damage, deaths, cooldowns used. The new Rotation tab shows what you actually pressed: your opener, cast frequency grouped into Rotation / Cooldowns / Utility, and a timeline of every button aligned to each pull. No one else in the ecosystem does this.
Source data
HOW WE DECIDE WHAT COUNTS AS AVOIDABLE
The "avoidable damage" and "critical kick" lists per dungeon aren't hand-curated by us. They come from a cross-log sampler that pulls the top 20 speed-ranked logs per dungeon and finds the abilities + casts that consistently appear across them. If top players take damage from it in 50%+ of runs, it's avoidable. If they kick the same cast in 50%+ of runs, it's critical.
That's it. No editorial curation, no favoritism toward any class or spec. When a new season's data rolls in, the lists update. When we add a new dungeon, we re-sample. The endpoint that generates these lists is the same one our scoring engine reads from.