How FiveRank measures

FiveRank reads the public FiveM server list. It is the same list the game's own server browser uses, so a server is on FiveRank because it is on that list, not because anyone applied or paid.

The server details (names, tags, locale, frameworks) are refreshed weekly. Live player counts are refreshed about every five minutes, and the server list page says how long ago the last refresh was.

The player rankings are ordered by players connected right now. When two servers have the same number, the one with more boost comes first. A country or region ranking is the same order, filtered to that country or region.

Nothing you can buy changes that order. Boosts have their own board, which is labelled as the boost board, and sponsored placements are labelled wherever they appear.

FiveRank records an hourly player reading for the 5,000 busiest servers and keeps 30 days of it. Servers an owner has claimed also keep one point per day for 365 days.

Peaks and averages are worked out from those readings only. History starts the day recording started for that server: nothing is backfilled, because FiveRank does not have data it did not record.

“Rising” compares a server's average over the last 7 days with the 7 before. Until two weeks have been recorded it compares a day with the day before, and says so. Only servers with enough readings in both periods and at least 25 players on average are compared.

Uptime is recorded for claimed servers. “Up” means the server was listed on the public server list at the time of a reading. It is not a statement about the server's own process: a server can be running and briefly missing from the list, or linger on it just after it stops.

A reading is only recorded when a refresh actually succeeded. If FiveRank itself was down, those minutes count as neither up nor down, so downtime FiveRank did not observe is never charged to a server. Each figure is shown with how much of the period it covers.

Once a month FiveRank sweeps the servers and counts which resources each one runs. A script's share is the servers running it out of the servers that sweep counted, 24,674 in the sweep of 1 November 2026. Every percentage is quoted against that number and never against the whole catalogue.

Only the 30 busiest servers running a script are listed for it, so a developer's reach can be a floor, not an exact count. When that is the case the figure is shown with a plus sign instead of passing for exact.

Movement in the script rankings needs two sweeps to compare. With one, there is no movement and none is shown.

A verified server is one whose owner has proved they run it: by putting a code into the server's own configuration, by a proof on the Cfx.re forum, or by a staff member checking. Nothing else earns the mark.

A verified developer has proved a GitHub or Cfx.re forum account is theirs, and the scripts and skills shown are read from that account. A claimed studio has proved its store is its own, by its own domain or by a person checking.

Paid plans and purchases change how a page looks and what analytics its owner sees. They never change a ranking, a count or a reading, and the player count, rank, uptime and review score always draw in the same style whatever skin a page wears, so they can be checked by eye.

Where money does buy visibility, it is labelled: a sponsored slot says so, and a paid “available for hire” listing is the placement that is paid for, not the verification, which stays free.

If a figure cannot be measured it is left out, not estimated. A server with too little history has no peak. A script with one sweep has no trend.

Found something that looks wrong? Tell us in the community Discord and say which page. A correction is faster than an argument.