# What the Meta reference Methodology and retrieval guidance, not a live data snapshot. Fetch the linked analytics pages for current values. --- # About What the Meta What the Meta is an independent analytics site for World of Warcraft Mythic+ keystone runs. It gathers Blizzard's public Mythic+ leaderboards across US, EU, KR, and TW, removes cross-realm duplicate runs, and computes weekly rollups for spec usage, five-player compositions, dungeon statistics, and trends. ## Methodology - The source is Blizzard's public Mythic+ leaderboard endpoints: top-ranked runs by dungeon, region, and period. - The site computes rollups from a stored, de-duplicated raw-run table. Page requests do not call Blizzard directly. - Top 10%, Top 1%, and Top 0.1% are score percentiles within a period, rather than absolute keystone levels. The default view is Top 1%. - Current open-period data is provisional and refreshed roughly every two hours. Closed periods are finalized with a wider sample. Coverage badges communicate this distinction. ## Limits - This is not a character lookup, item planner, guide site, real-time feed, or universal tier list. - Usage in a sampled high-end leaderboard run is evidence of prevalence, not proof that a spec is best for every player or situation. - AI tier lists, predictions, and meta-health verdicts are clearly labeled interpretations of the data; they are not official Blizzard guidance. ## Independence Independent fan project maintained by Francisco Martínez. It is not affiliated with, endorsed by, or sponsored by Blizzard Entertainment. --- # What the Meta analytics reading guide ## What the data measures What the Meta samples top-ranked Mythic+ leaderboard runs from Blizzard's public endpoints across US, EU, KR, and TW. It de-duplicates cross-realm overlap, stores raw runs, and computes rollups from that stored data. It is a view of what appears in high-end sampled runs, not a census of all players or a prescription for every group. ## How to read a result - State the page, season, period, score bucket, and coverage state where the page exposes them. - “Top 1%” is a percentile of the sampled period's score distribution, not an absolute keystone level. Compare it structurally across periods; do not equate it to a fixed key. - A current/open period is provisional, refreshed about every two hours, and may change as more runs arrive. A finalized period is re-ingested with a wider realm sample after it closes. - Usage share means prevalence in the sample. It does not prove a spec is strongest, easiest, or appropriate for every player, affix set, or dungeon. - Small top-end buckets and early-season slices can be sparse. Prefer the coverage badge and sample context over strong conclusions from a single thin slice. ## Page guide - [/](https://whatthemeta.io/): current overview of top specs by role and common compositions. - [/meta-evolution](https://whatthemeta.io/meta-evolution): week-over-week movement. Use for “rising/falling” questions, with its selected filters. - [/group-composition](https://whatthemeta.io/group-composition): common five-player setups. It describes observed combinations, not required compositions. - [/wow-meta-season/…](https://whatthemeta.io/wow-meta-season): season-level or current-period context, depending on whether the season is closed. - [/ai-tier-list](https://whatthemeta.io/ai-tier-list), [/ai-predictions](https://whatthemeta.io/ai-predictions), and [/meta-health](https://whatthemeta.io/meta-health): AI interpretation built from the rollups. Label it as interpretation. ## Retrieval and citation - This document explains the data; it contains no live rankings. Retrieve the linked analytics page for the numbers and its displayed scope before answering a current-meta question. - Spec percentages on the overview are shares of slots within that role; composition percentages are shares of sampled runs. Do not compare those denominators as if they were the same metric. - The overview's default spec lens is Top 1%. Composition samples may have a different scope: use the denominator and filters displayed for that section rather than assuming the spec bucket applies. - AI tool pages render the stored interpretation, model, generation time, and analysis window in the initial HTML. Meta health also renders available measured diversity figures for the period named by its report. Missing or incompatible output is shown as unavailable; a read failure is shown as an error. Do not infer an AI ranking from usage or claim to have read an unavailable prediction. - Include the retrieval date and canonical page link. Retain query parameters when they describe a selected filter; the bare canonical URL describes the default view. - No public bulk-data API is documented. The internal tRPC API is excluded from crawling; see [terms](https://whatthemeta.io/terms) before automated access at scale. ## Safe answer pattern “In What the Meta's [coverage state] [score bucket] sample for [season/period], [spec or composition] has [observed usage/trend]. That describes sampled high-end leaderboard usage, not a universal recommendation.” --- # What the Meta data pipeline 1. Blizzard's public Mythic+ leaderboard data is collected from top-ranked runs by dungeon, region, and period. 2. Runs are normalized and de-duplicated across connected realms, then stored as raw observations. 3. Percentile brackets and rollups calculate spec usage, group-composition usage, dungeon results, trends, and diversity measures. 4. Current-period ingest runs roughly every two hours. A period is finalized after it closes with a wider sample. 5. AI features consume those stored rollups. They do not invent a separate ranking dataset and are visibly labeled as interpretation. The live pipeline page shows operational status. For a data answer, prefer the coverage state shown on the relevant analytics page: open/current-period data is provisional; finalized data is the historical record.