Who Actually Plays PUBG on PC Now? 2,207 Randomly Sampled Matches Map the Real Market

How this was made (method & sample) Source: the PUBG official API's random match-sampling endpoint, /shards/steam/samples, across three 24-hour windows (2026-08-29, 08-25, 08-22), deduplicated to 2,207 matches. Each match was then pulled via /matches/{id} for its mode, map, start time and every participant's ID. After excluding custom rooms, tutorials, the training range and casual bot matches, the analysis set is 1,430 public BR matches totalling 93,106 human participant records. Humans and bots are not guessed apart: in the official data every bot's playerId begins with ai.. Captured 2026-08-30, current season division.bro.official.pc-2018-42. No account is named and no individual ranking is published.

Almost every data article on this site draws its sample from a leaderboard, because that's the one reliable way to get a group of strong players. But leaderboards carry a fatal sampling bias: they only know the top of the pyramid. What the top plays, when it plays, and how clean its lobbies are can look nothing like the game as a whole.

So this time we inverted the question: don't pick the players — let the official API throw us a random batch of matches. PUBG exposes a rarely used endpoint that returns a random sample of matches from a 24-hour window. We pulled three windows, took every match apart, and went after a basic question almost nobody has answered with data — who is actually playing PUBG PC right now, and what are they playing?

First finding: TPP vs FPP isn't 50/50, it's 90/10

Adding up every human participant across 1,430 public BR matches, third person (TPP) takes 87.5% and first person (FPP) just 12.5%.

If you live in the FPP world — streams, esports, forums — that number probably stings. It is entirely logical though: FPP is the taste of people who play seriously, and serious players are always the minority.

More interesting still, the ratio depends strongly on party size:

The closer you get to solo, the more weight FPP carries — solo's FPP share is nearly 3× squad's. But mind the denominator: solo is only 2.4% of all play, so "FPP is more popular in solo" describes a relationship inside an already tiny pool.

In absolute terms, squad TPP alone accounts for 74.2% of all human play. This game has exactly one mainstream mode, and its lead is too large to call a contest.

Ranked is a different world

Split by matchType, ranked (competitive) is 22.2% of human play; normal mode is 77.8%.

FPP's share inside ranked is 16.6%, higher than normal mode's 11.4% — people who play seriously do lean FPP. But even in ranked, TPP still leads 83/17. Ranked's mode distribution is also more concentrated:

Across three 24-hour windows, not one ranked solo-FPP match was sampled. That doesn't mean it never runs — it means the queue is small enough that random sampling misses it.

In a normal-mode lobby, 30% aren't people

This is the most immediately useful finding of the whole sample, and the one nobody talks about.

Because the official data marks a bot's playerId with an ai. prefix, "how many humans are in this match" is something you can count directly — no estimating, no guessing. The answer: 30.1% of a normal-mode lobby is bots, and the emptier the queue, the more absurd the ratio.

"Humans per match" is the most honest health indicator a queue has: squad TPP fits 75 real people into a lobby, solo-FPP only 18 — the rest of the slots are filled by bots. Normal-mode solo is, in substance, already a game against AI.

This contaminates your stats directly. Judging yourself or anyone else on normal-mode K/D, headshot rate or survival time means computing against 70%+ bots in solo, 80% in solo-FPP. To compare skill, only ranked data is clean; normal-mode figures need a label, or at minimum to be read separately — which is exactly why this site keeps the two apart.

Region: the match data has none, but the clock talks

We wanted to compute per-region populations directly and hit a wall: PUBG match data has no region field. More precisely, that random-sampling endpoint returns the same list no matter which shard you ask (pc-as, pc-na, steam…) — it is a platform-wide sample, not a regional one.

So region can only be inferred from start time. Spreading the 1,430 public BR matches across UTC hours:

The peak lands at 13:00–14:00 UTC — 22:00–23:00 in Korea and Japan, 21:00–22:00 in China. 73.9% of matches fall inside 06:00–17:00 UTC, Asia's afternoon through late night.

But the persuasive part isn't the traffic curve, it's the second column: the FPP share swings violently with the clock. The low is 3.7% at 07:00 UTC (Asian evening); the high is 45.9% at 23:00 UTC (European night). The quietest block carries more than four times the FPP weight of the busiest one.

The inference chain is clean: if FPP preference were uniform worldwide, it wouldn't move with the clock. It moves, regularly, which means different time zones seat two populations with different taste. TPP is Asia's mainstream; FPP is the West's relative mainstream — the West is simply much smaller, so the global total flattens it to 12.5%.

Which regions are still alive

There's one more indirect but hard piece of evidence. The official leaderboard endpoint only answers when a region actually has a ranked environment, so we probed all 11 PC shards this season:

This lines up exactly with the region merge in Update 32.1 (October 2024): on PC, AS + KR/JP → AS and OC + SEA → SEA. Japan, Korea/Japan and Oceania have produced no leaderboard since.

The official direction pushes ranked toward TPP — the opposite of demand

To align with PUBG Esports going fully third-person in 2026, Update 40.1 changed which perspectives ranked offers where: RU gained Duo and Squad TPP ranked, NA and SA gained Squad TPP, and NA and SA lost Squad FPP ranked.

Set that beside the clock data and it reads as irony: FPP's real demand is concentrated in exactly the Western hours, and what got removed is the relatively popular mode inside the minority regions. From the global aggregate it is defensible (FPP is 12.5% overall); from the regional lived experience it hurts — which probably explains why complaints about changes like this cluster so tightly in the North American community.

Language: this is a Chinese and Russian game

The API exposes no country data, but Steam review counts can be queried per language, and they're the closest public proxy for national distribution. Taking the last 365 days of reviews (171,053 in total) represents "who plays now" better than the all-time figures:

Two trends stand out: the Russian-speaking share has doubled (12.5% → 25.8%) and English is shrinking (16.4% → 11.9%). That matches the clock curve — the 18:00–23:00 UTC block, high on FPP and low on volume, is mostly Europe and Russia, not North America.

One more caveat: Korea also has a separate Kakao client (pc-kakao still carries a full leaderboard), and none of it appears in the Steam sample or in Steam reviews, so Korea's true share is understated here.

Maps: three of them take 80%

Taego, Erangel and Miramar total 79.6%. The 8×8 maps are the outright mainstream; the small ones are garnish. Map rotation moves weekly results, so read this as a late-August 2026 snapshot rather than a long-run average.

Being honest: what this didn't verify

• "Human participation" is not "number of players." One person appearing in 5 sampled matches counts 5 times, so every share here is a share of play, not a share of people. Heavier players carry more weight — usually what you want when asking where a market is heading, but it cannot answer "how many people play FPP."
• The sample covers the Steam platform only. The samples endpoint returns the same list for every shard, so it is a platform-wide sample; the Kakao (Korea) client and the console versions are out of scope.
• It assumes sampling is roughly uniform across the 24 hours. The official docs don't describe the sampling method, and the entire clock analysis rests on that assumption. The three windows (Friday, Tuesday, Friday) agree with each other, but this remains the weakest link.
• The clock is not the region. Inferring region from UTC hour mixes Asian night owls with European office workers. It supports conclusions at the scale of "Asia's volume vastly exceeds the West's"; it cannot produce precise per-region shares.
• Steam review language is neither nationality nor player count. It is a proxy for who speaks up, and it inflates whichever language group is currently agitated about a review-bomb.
• Only three days. 2,207 matches across three 24-hour windows is plenty for a stable structure like mode distribution, and not enough for map rotation or event cycles that vary week to week.

Conclusion

Put six numbers side by side and PUBG PC's shape today is unmistakable: this is an Asian, third-person, squad game. TPP 87.5%, squads 82.3%, 73.9% of play in Asian hours — three independent indicators pointing the same way, none of them by a narrow margin.

And for anyone who wants to read their own data seriously, the number to remember is that 30.1%: a normal-mode lobby is a third bots, and solo is seven to eight tenths. A good chunk of the pretty numbers you post in normal mode were handed to you by AI. Ranked is the honest mirror.

Want to know how much of your own numbers are real? Look yourself up on L!NCHPIN and read ranked and normal mode separately — the gap between the same player's two sides is usually wider than you'd guess.

Data is a single snapshot taken 2026-08-30, sampled from three official 24-hour windows. It does not represent a long-run average or other platforms. No account is named and no individual ranking or account information is disclosed.