Ok, final thoughts on our last month of aggregate data before Statshark stops publishing it.
A caveat: something I’ve known for a while now is that Statshark was adjusting the values of some of this global user data to protect the source (I suspect that was limited to the stolen data, not anything else about the site, which the owners could and still can get legitimately for now, such as the individual vehicle data, I don’t believe any of that is being altered). Since we now all know the source was Pluspy, I’m less inclined to keep that confidence. I don’t know what factor he applied to the data to mask it or how to reverse engineer it either, so we have what we have. This is, of course, the problem with all stolen data: you can’t really compare with the original. But to the degree I (and you) are still inclined to trust any of it, here’s what September (and the eight months we could see inside actual proprietal data about War Thunder) tells us.
1. Steam data was a pretty good predictor
Steam users are only a portion of the player base, and people often say you shouldn’t use them to discuss how War Thunder is doing. This is true, but the decline in player sessions in Statshark over the last half-year and the decline in Steam numbers actually pretty closely tracked the whole time (I’m using steamcharts.com for this, other third-party Steam trackers have slightly different values).
Spoiler
The exception there was February, Statshark’s first attempt, which is an outlier, can’t explain that one. But the rest is so close it’s fair to say Steam data has some use (and given that’s all we’ll have again after this month, it’s interesting for these last 7-8 months it has tracked the all-player patterns very well).
2. Vehicle events don’t change play patterns much
After eight months, it’s safe to say that the single vehicle events do not change relative play patterns very much, with a couple exceptions. If we look at the months with more than 7 days of a single-vehicle event, like the F-106 event, we could expect that players would play more of that event mode and less of the other modes. What we see is a swing of about 4% of ground mode sessions or 8% of air mode sessions back and forth, or about 4 million user sessions per month total (out of 220 million+ in a month), as the “swing vote.” This is not a huge effect, and suggests most people kept playing the mode they like regardless of the single-vehicle events this year. Also notable is the gains are almost all on the realistic side (ignoring SB here because those “user sessions” are not really comparable), over arcade. If anything, arcade play in ground and air goes down during vehicle events (probably because of players chasing multipliers).
Spoiler
The exception is naval… which a naval event will lead to a significant swing to that mode (about 30% more player sessions in both AB and RB this year), that’s maybe 1.5 million user sessions attracted away from the other modes, or about a third of the swing between the other two modes. While it’s still a big deal for naval popularity when a vehicle event happens, basically even most of the people who like these vehicle events don’t change their play for the naval ones.
3. Air modes have been declining in popularity relative to ground this year
In a time when the whole pie was shrinking (mid-2025 was the worst mid-year in terms of player loss proportionately in War Thunder’s history, both Steam and Statshark agree, with a 20% player loss March-Sept compared with 10% in 2023, their previous worst year), air mode user sessions shrunk significantly faster than ground modes.
Spoiler
Between March and September, ground modes (AB/RB) fell about 15%. But the air modes combined fell over 30%, with ARB falling faster than AAB. Ground, and specifically ground RB, with over 50% of all AB/RB user sessions, is the majority mode now. This is what makes it so baffling to me that there have been no ground events at all since June. Maybe Gaijin thought it was healthy enough and wanted to use this midyear to prop up the other modes more, with two naval events either side of the F-106 event?
4. Naval changes made no major difference to player patterns
A series of naval changes this year, and a major update focussed on naval, succeeded in raising the player base of naval somewhat (from about 1.4% of player sessions in Feb/Mar to about 2.3% in September… but the 30% boost mentioned above from the Mackesen event running would have provided a lot of that in September too, though so any overall gain independent of events could have been quite small; but at least with the Statshark sessional stats stopping this month naval fans are going out on a high note, scraping in above 2% of sessions like that, though), but the overall proportion between NAB and NRB player sessions ended up in September (67.8% AB) almost exactly where it was in February (67.6% AB).
Basically both sides of the ongoing debate in this forum appear to have been wrong. The naval changes to (and to AB in particular) did not bring a whole lot of new players to naval overall, as many as hoped (although they did bring some). But people who said the naval AB changes would drive people from that mode also haven’t seen their predictions bear out. At the time I said the NAB changes weren’t drastic ENOUGH and that the two modes needed to be much more strongly differentiated for them to stay independently healthy, even if it meant I likely wouldn’t play one of them anymore: I’m still comfortable with that assessment.
The real problem with naval has persisted past decompression, that unlike the other modes with different AB/RB BR patterns that give both submodes a distinct role, NRB and NAB are still both played basically in the same proportion at all BRs, essentially in direct competition with each other:
Spoiler
The exception there may be NRB at 3.0 and below, which has declined to almost nothing now. Seeing a lot of people reporting 1v1 and even 1v0 coastal games now (seen a couple myself). There the remaining players who don’t want to just fight bots have picked a mode, for better or for worse. These things have a “rolling downhill” effect, as people abandon the 6 minute queue times and increasing bottedness of it all, so at 3.0 and below it’s safe to say NRB is a dead mode. NRB remains healthy at all the bluewater BRs though, at least for now.
5. The US is still a top 3 nation only because of the air modes
Players by country haven’t changed much over six months, China continues to develop slowly into the #4, firmly passing UK and Japan this year, Italy and Israel continue to need the most love. And the US ground game continues to be where it’s weakest in terms of player base, whereas in the air modes it continues to be the dominant nation by a significant margin. (This should be concerning for US mains, though, if air continues to decline as a mode relative to ground.) One can’t really help but attribute this to high-tier/end-game problems in ground RB, given that it’s the majority mode, which really seem to have broken US viability recently relative to Germany or USSR.
Spoiler
Naval is not hugely relevant for any nation, although the strongest there is probably Japan, with 6% of its AB/RB user sessions being in naval games. Also notable here are the relative success within their modes of French air and Italian ground overall, compared to the other minor nations (one has to wonder if the recent addition of subtrees (which have mostly been on the air side of France so far) could have had something to do with that. Some good news there for Italian ground mains anyway.
