Methodology · v3.2
No black box. We show the working.
Every server earns a single 0–100 composite score from five components, each measured from real data and combined with fixed weights. The same weights below drive the score on every listing — nothing is hand-tuned per server, and featured placements never change where a server ranks.
Member count on a logarithmic scale, so a community of a few thousand can still rank well against the giants. Big numbers help, but they never dominate on their own.
The share of members online and talking. A smaller server with a buzzing channel beats a huge, silent one.
Week-over-week member change relative to size. Servers picking up steam get a lift; ones bleeding members slide.
Discord's own signals — Verified, Partner, and Community status — as a proxy for legitimacy and moderation.
How squarely a server sits in its niche, judged on its sub-category and tags. A focused server ranks above a vague catch-all.
output: weighted sum of the five components, normalised to a single 0–100 score (one decimal).
How the score comes together.
Each component is scored 0–100 on its own, then multiplied by its weight and averaged. Size carries the most weight (×2.0) because scale genuinely matters, but a big server that's quiet, shrinking, or unverified won't hold the top spot against a smaller one that's alive and growing.
Scores refresh roughly every 15 minutes as new data syncs. We keep a longitudinal record of every server's members, online count and score over time — that history powers the growth charts and the “tracked since” figures on each server page.
Featured placements
Paid spots are labelled — and never move the rankings.
Sponsored and featured servers are always clearly marked as such. They appear in dedicated featured slots but their organic rank is calculated by exactly the same composite as everyone else. We may earn a commission when you join some servers through our links — see our affiliate disclosure.