Session Fill Rate
Session fill rate is the share of available player slots actually occupied across your running game servers, averaged over each session's full lifespan. A server built for ten players holding four is running at forty percent. It is the number that connects matchmaking behaviour directly to hosting cost.
Also called
fill rate, game session fill rate, matchmaking fill rate
,
How Do You Measure Session Fill Rate?
Session fill rate = occupied seat-minutes ÷ available seat-minutes
It's measured over the whole session, not at match start. A server that starts with 10 players and finishes with 6 was full for a moment and part-empty for the rest of a match billed in full. Each part of the formula hides a choice:
Occupied seat-minutes. The minutes each seat holds a connected player. A seat held for a party member who is still loading usually counts. Bots are a separate question, covered below.
Available seat-minutes. The seats the mode was designed for, not the most the server could technically hold, multiplied by how long the server ran. Minutes before the first player connects and after the last one leaves count too, because they are billed. On a multi-room server, count each room.
The window. Both sums are added across many sessions, then read per region, per queue and per hour of the day. One global average can hide a quiet region, an off-peak hour or a niche mode.
Measured this way, fill rate can be compared week to week. A snapshot at match start mostly shows how the start rules are set.
Where in a Session Does Fill Rate Drop?
Fill rate is decided at three moments in a session's life, and each has its own cause.
At the start. A matchmaker that starts a match at its minimum player count, or when a wait timer runs out, opens a server with seats already empty. Both are reasonable rules for keeping queues short. Both lower fill from the first minute.
During the match. Players leave: they disconnect, quit, or finish their part early. In a Battle Royale, elimination empties seats by design, so fill falls through every match even when nobody quits. Backfill can refill seats in modes where joining late is acceptable.
At the end. The match is over but the server is still running: a results screen, a slow shutdown, or a server waiting to be told to stop. Every one of those minutes is capacity with nobody in it. Stopping the server as soon as the last player leaves is often one of the easiest gains, because no player waits longer for it.
Which moment weighs most differs from game to game, and so does how much of each loss is worth fixing. A studio that measures fill at all three can see where its empty seats come from, then choose which to accept and which to work on.
Why Do Some Game Genres Fill Less Than Others?
Before any tuning, a game's design sets roughly how full its servers can run.
Lobby size. Filling 2 seats takes seconds. Filling 100 takes longer, and every second waited is a reason to start with fewer.
Match length. A long match has more time to lose players. A short one spends a larger share of its billed time in warmup and shutdown.
Drop-off by design. Elimination modes empty seats on purpose, so their average fill sits well under their starting count. Extraction games are well known examples of matches whose fill rate decays over time.
Split pools. Separate queues for each platform, input or mode each fill from fewer players.
So comparing fill across genres says little. Comparing a game's own queues, regions and hours says much more.
Within a game, fill trades against wait time. Waiting for a fuller lobby raises fill and lengthens queues. That's why fill is best read next to the ticket expiration rate: if both rise together, players are timing out before they match. Our guide to optimizing session fill rate covers the matchmaker rules that move both.
Do Bots Improve Session Fill Rate?
Bots fill a lobby. Which may be excellent to ensure a perception of a fuller queue size and thus, player satisfaction for your game. However, they don't fill it in the sense that matters for cost.
Bots have real design value: they let a mode start sooner at quiet hours and can give new players a gentler start. But the server still runs at full-lobby size with fewer human players, and the bots add costs of their own. Our analysis of fill rate and hosting costs describes three layers:
The seat a human didn't take still lowers fill.
Server-side AI adds CPU load.
Each bot's state is sent to every client, adding egress that grows with player count and tick rate.
When bot seats are counted as empty, fill rate shows the real gap, and the cost of bots appears as its own line in the hosting budget rather than as unexplained overspend.
A word from our sponsor (ourselves!)
Every empty player slot is capacity you pay for and nobody uses. One studio on Edgegap's matchmaker averaged 91% of player slots filled across its sessions in Q2 2025, so fewer servers carried the same players.
Edgegap's Take (just our opinion, take it with a grain of salt!)
Aim High, Then Know Where and Why You Settle
Best practice points to aiming high. Our fill rate calculator models optimized matchmaking at 95%, and one studio on our matchmaker averaged a 91% session fill rate in Q2 2025 (platform data), a single title rather than a platform average. High fill is reachable on real traffic.
A live game rarely holds its best number everywhere. A quiet region, an off-peak hour or a niche mode will often sit lower, and that can be fine. What matters is knowing where and why, then deciding whether it's acceptable or worth more tuning.
A place to start: each week, read time-weighted fill per region and per queue next to its ticket expiration rate. Low fill with few expired tickets usually points to start rules that are looser than they need to be. Low fill with many expired tickets points to a pool that's too small, where a separate low-population profile or a shared regional pool tends to help more than stricter rules.
,
