The latest Steam Survey data became available this week, showing a record number of headsets in use on the platform as well as a surprising shift in the popularity of specific headsets.

That’s based on March 2020 data released this week from Valve’s Steam Survey. Each month the company collects info from Steam users to determine some baseline statistics about what kind of hardware and software is used by the platform’s population, and to see how things are changing over time.

This latest data is the first time we’re seeing the numbers since Valve revamped the Survey to collect more accurate information about which headsets are being used on Steam. In addition to revealing new headsets in the data for the first time—like Quest, Pimax, and PSVR—it also understandably brought a significant shift in the numbers from the prior month.

While the Steam Survey had previously only detected headsets that were connected to a user’s PC at the time of data submission, the new method stores information about any headsets connected to the PC in the last month and reports those headsets accordingly.

As the data represents the number of headsets connected to Steam over a given month, we call the resulting figure ‘monthly-connected headsets’ for clarity.

Monthly-connected VR Headsets on Steam

The latest figures from March show that 1.29% of Steam users had connected VR headsets to their PCs over the course of the collection period. That’s an increase of 0.28% over the previous month, but there’s two caveats here: first, we can’t be sure how much of that is attributed to actual growth vs. the change in data collection methodology, and second, Valve advised that most of the Survey data is collected in the first few days of each month, meaning that we won’t see most of the impact from the release of Half-Life: Alyx (which came later in the month) until the next set of survey data.

The data shows a bit of chaos surrounding the change in collection methodology, but either way, it’s the next data point in a trend which has been stepping upward on average. To put the number into perspective, the latest Survey data also shows that 0.87% of Steam’s population is using Linux and 3.80% is using MacOS.

While Valve’s data is a useful way see which headsets are most popular on Steam, the trend of monthly-connected headsets has always been obfuscated because the data points are exclusively given as percentages relative to Steam’s population—which itself is an unstated and constantly fluctuating figure.

To demystify the data, Road to VR has created a model based on the historical data, along with official data points directly from Valve and Steam, which corrects for Steam’s changing population to estimate the actual count—not the percent—of headsets being used on Steam.

While the percent of Steam users using VR headsets didn’t reach its highest point in March, our model estimates that March saw the greatest count of monthly-connected headsets to date.

With more accurate information from Valve, our latest estimate of the number of monthly-connected headsets on Steam stands at 1.77 million. That’s a sizeable leap, and though we can’t be sure how much of the change is from real growth rather than the change in collection methodology, it’s a record high number for monthly-connected headsets on the platform. If the current trend continues, we expect to see some 3 million monthly-connected headsets on Steam around January 2021.

Share of VR Headsets on Steam

As for the share of individual headsets on Steam, things got understandably jostled around, including the introduction of headsets previously not shown in the Survey. Valve says the changes offer a more accurate picture of headsets in use on the platform, so here’s what we now know:

  • Valve’s Index has surpassed all WMR headsets in use on Steam (10.94% vs. 8.28%)
  • Oculus Quest has surpassed Vive Cosmos (2.89% vs. 1.25%)
  • Oculus Rift S is the single most popular headset in use on Steam, surpassing the original HTC Vive (27.05% vs. 26.67%)
  • Oculus Rift (CV1) was overrepresented in earlier data; collectively all consumer Oculus headsets now account for 45.50% on Steam (53.73% previously)

Lesser-used headsets are also now showing in the data. Coming in at less than 1% of the share of headsets on Steam is Pimax 5K Plus, Pimax 8K, and even Playstation VR (which is not officially compatible with Steam but enabled via some third-party workarounds). There’s also 2.91% of “other” headsets shown, presumably a range of lesser-known and likely even misidentified headsets.

SEE ALSO
2019 Was a Major Inflection Point for VR—Here's the Proof

As ever, it’s worth noting that the Steam Survey only gives us a glimpse of the overall VR market, as it only counts headsets connected to Steam. That means it doesn’t count some portion of Rift users which may not use Steam at all, nor other major headsets like PlayStation VR on PS4 and Quest standalone.

Newsletter graphic

This article may contain affiliate links. If you click an affiliate link and buy a product we may receive a small commission which helps support the publication. More information.


Ben Lang
Ben is the world's most senior professional analyst solely dedicated to the XR industry, having founded Road to VR in 2011—a year before the Oculus Kickstarter sparked a resurgence that led to the modern XR landscape. He has authored more than 3,000 articles chronicling the evolution of the XR industry over more than a decade. With that unique perspective, Ben has been consistently recognized as one of the most influential voices in XR, giving keynotes and joining panel and podcast discussions at key industry events. He is a self-described "journalist and analyst, not evangelist."
  • Xron

    Now next step is 10kk+!

  • A VR Enthusiast

    Oh yeah, and we are just starting.

  • impurekind

    Good stuff. Hopefully that number just keeps on rising–and I see no reason why it shouldn’t.

    • Xron

      Supply chain atrm.

      • jimmy

        ik if oculus was selling a vr headset to anyone who wanted to buy right now they would have probably sold one million extra hmd in the past 3 months

  • Matteo Valles

    That trend line makes it look like we are about to be at the elbow of an exponential curve of VR usage…hope that’s the case

  • mepy

    2017, 250.000 increase
    2018, 500.000 increase
    2019, 1 million increase

    So, 2020, 2 million increase?

    • Jistuce

      One can but hope.

  • The usual informative article, Ben. But can we just stop and think of those poor people still using the DK1 and DK2? :D

    • Andrew Jakobs

      Why? if they’re happy with using that one, who are we to feel sorry for them?

  • Andrew Jakobs

    Yesterday was the first time in many years that I actually got a screen popup with a steam hardware survey (and it detected my vive pro even though I hadn’t turned on).

  • Andrew Jakobs

    yeah, you’re not the only one, me too.

    • Tarik

      So does this mean thing the results could be inaccurate because not everyone has taken the survey

  • Cragheart

    2 million this year, 32 million in 2024?

    PC performance also used to double every year, but later slowed down.

  • DickDastardly

    “Valve says the changes offer a more accurate picture of headsets in use on the platform”

    This is technically true, so long as you understand that “the platform” is Steam VR. But this is not the same thing as claiming that the new survey methodology gives a more accurate picture of headset use in general (spoiler alert: it doesn’t).

    Previously headsets were counted if connected to the PC at the time the survey ran. This underestimated the actual number of headsets in use because some people don’t leave their headset permanently connected, but the relative proportions of PCVR native headsets would generally have been correct as there’s no reason to suppose users of any particular headset are more or less likely to do this. (PSVR and Quest users are obvious exceptions as their users aren’t reliant on PCVR games so they will have been under-represented in the old data).

    The last semi-reliable survey data we have from before the change showed that Oculus HMDs were now in first and second place amongst Steam users (having outsold all other HMDs combined, and the Index in particular by at least 7.5:1). Coincidentally, as soon as this became clear, Valve decided to change the survey methodology.

    The new methodology, counts any headsets which have been used in Steam VR in the last month (and only counts currently connected headsets if no Steam VR usage is found).

    This change had the effect of massively distorting the data because headsets which rely on Steam VR exclusively (like the Index) are now far more likely to be counted (as they’ll now show up in the data if used even once in the last month, whether or not they’re connected at the time of the survey).

    Meanwhile, headsets whose primary store isn’t Steam (like the Rift and Rfit S) are now relatively far less likely to be counted (as a Rift user might have played Oculus store games every day, and perhaps even Steam games which offer an Oculus API option, without ever showing up in the Steam VR logs, so they’ll still only be counted if they happen to have their headset connected when the survey runs).

    Far from CV1 users being over-represented in earlier data (as the article claims without any explanation or justification whatsoever), Oculus users are now massively under-represented in the new data, making it almost entirely useless as a means of judging the relative success of headsets.

    If you think I’m quibbling over a small technical detail then you only need to look at the enormous size of the effect which the change to the methodology had: The proportion of Rift CV1 users counted immediately fell by a whopping 42% (their biggest change since Valve first included VR in the survey), whilst DK1 and DK2 users declined by 88% and 26% respectively!

    In comparison, OG Vive users actually went up, so it makes no sense to claim the changes are just due to users of old headsets upgrading, unless you can explain why hundreds of thousands of Rift users all decided to do so (at the exact time Valve changed the methodology) whilst no Vive users did (and why none of those now former Rift users sold or gave away their old headset).

    TLDR It turns out Valve’s commitment to openness and transparency only held until it became sufficiently embarrassing for them, at which point they fudged the data to inflate the figures for their headset at the expense of their main rival’s. As a result, the Steam survey data no longer even gives an accurate picture of the proportions of headsets amongst those who use Steam, much less of the broader VR market.

    • benz145

      I think you may be making some assumptions which aren’t correct.

      My understanding (from what Valve has shared with us) is that if SteamVR doesn’t see any headset being connected in the last month, then the Survey will fall back to the old method by checking for devices via USB. I’m not sure how this biases one particular headset over another.

      • DickDastardly

        Quoting my response to an earlier article:

        “Consider, for example, a user who only has their headset connected to their PC on average one day a month. With the old methodology they only had a 1 in 30 chance of having their headset connected and detected when the survey ran.

        But compare that to their chances with the new system: If they exclusively use Steam VR (let’s say because they’re an Index owner) then they’re now guaranteed to be counted in the survey because they’ll show up in the Steam VR logs for the month, whether or not they happen to have the headset plugged in at the time of the survey. In other words the chance of their headset being counted has gone up by a factor of 30!

        If, however, they’re a Rift owner who has Steam installed for pancake games but only uses the Oculus store for VR content, then the chances of their headset being counted hasn’t changed – it’s still only 1 in 30. In fact this may even also apply if they play VR content on Steam which uses the Oculus API rather than Steam VR (as presumably they wouldn’t show up in the Steam VR logs then either and so would still only be counted if their headset happened to be plugged in at the time of the survey).

        Of course, this is just an example – I expect most VR users have their headsets plugged in for far more than one day a month, but it explains why there was such a large drop in CV1 users at exactly the time the methodology changed whilst OG Vive users actually went up.”

        In other words, the new methodology makes headsets which rely on Steam VR far more likely to be counted than before, whereas those who avoid Steam VR (by buying games on the Oculus store or even by buying Steam games which use the Oculus API) are no more likely to be counted than they were previously. Hence the massive change in the relative proportions reported.

        • benz145

          Your main argument seems to be that PC VR headsets used with Steam are more likely to be counted under the new methodology than PC VR headsets not used with Steam.

          If we grant that everything you said is correct, I don’t see this as being a problem; if the point of the Steam Survey is to see which headsets are being used on Steam, then a change that makes the survey less likely to count headsets not used on Steam makes the Survey more accurate by definition.

          • DickDastardly

            It’s not just that the change made the survey results less representative of headsets which aren’t used with Steam (which it did), but also that it’s now less representative of headsets which are used on Steam but which don’t use Steam VR.

            Given that a significant proportion of Steam games offer the option to use the Oculus API instead of Steam VR and that pretty much 100% of Oculus users are going to take advantage of that option for better performance, this means that the results are now even more heavily skewed against Oculus headsets and in favour of those which require Steam VR (like the Index).

            Again, just to stress the size of the effect, Rift CV1 users counted fell by 42%! Unless you think 42% of CV1 users all had their cable die immediately before Valve changed the methodology, then that’s approximately the size of the distortion in the data which the change caused.

            Does this matter? I’d say it does – after all the only reason RoadtoVR and other outlets report on the survey at all is because it was the best publicly available proxy we had for the success of VR in general and of the relative success of particular headsets. The fact that 100% of outlets which reported on the change erroneously claimed that it’s now more, rather than less, accurate is problematic, to say the least.

          • benz145

            I can’t speak for others, but we have never looked at the Steam Survey as a marker for success of all PC VR headsets, only those that are used with Steam. That’s made pretty clear in our reporting, like in this article:

            As ever, it’s worth noting that the Steam Survey only gives us a glimpse of the overall VR market, as it only counts headsets connected to Steam. That means it doesn’t count some portion of Rift users which may not use Steam at all, nor other major headsets like PlayStation VR on PS4 and Quest standalone.

            In my experience, the vast minority of VR games on Steam do not offer use of the Oculus API, so I would expect that the difference in the count of Oculus headsets which ‘play VR games on Steam but exclusively those without SteamVR’ is very small. You seem to be assuming that niche intersection of VR users is 100% of what led to the change in the Rift’s numbers, but haven’t considered other factors which I’m willing to bet are a much more significant contributor.

            The Steam Survey may not be 100% accurate, but I do believe it’s more accurate now than it was before based on the changes Valve made.

            You seem to suggest that Valve designed the new Survey methodology in their favor, but, in the absence of any evidence for that, Occam’s Razor would tell us that if that was their intent, they would have done that long before Oculus headsets had become the plurality of headsets used on the platform, which happened back in 2018 (https://www.roadtovr.com/oculus-rift-takes-lead-htc-vive-steam-majority-market-share/).

            Having spoken with many of the VR folks at Valve over the years and understanding the way they work, I do think their goal with the Steam Survey (overall, not just the VR portion), is to give developers the most accurate glimpse of the Steam population as they can so that developers can make informed decisions. I do not see evidence to suggest their changes were to boost or reduce the numbers of any particular headsets.

          • DickDastardly

            In my experience, the vast minority of VR games on Steam do not offer use of the Oculus API, so I would expect that the difference in the count of Oculus headsets which ‘play VR games on Steam but exclusively those without SteamVR’ is very small. You seem to be assuming that niche intersection of VR users is 100% of what led to the change in the Rift’s numbers, but haven’t considered other factors which I’m willing to bet are a much more significant contributor.

            In my own experience, it’s very common for Steam games to offer an option to use the Oculus API (just off the top of my head in my favourite genres of shooters and driving sims: Contractor$, Pavlov, Onward, Alien Isolation, GTA V, Assetto Corsa, Dirt Rally 1 & 2, KartKraft, Projects Cars 1 & 2, and Elite Dangerous all offer the choice).

            In any case, there has to be some explanation for why CV1 users fell by 42% whilst OG Vive users actually rose. Naturally you’d expect there to be some decline in the proportion of older headsets used as people upgrade, but for reference the decline in CV1 users in the previous month was only 7% vs the month before. I’m entirely agnostic as to the proportions of what makes up the huge additional drop we saw when Valve changed the methodology – it could be primarily due to to the survey now under counting Rift users who use Steam exclusively for pancake games, or it could be due to under counting Rift users who do use Steam for VR but with the Oculus API. Do you have any other explanations? (bearing in mind that it’s the single largest change in the data we’ve seen since the survey started).

            You seem to suggest that Valve designed the new Survey methodology in their favor, but, in the absence of any evidence for that, Occam’s Razor would tell us that if that was their intent, they would have done that long before Oculus headsets had become the plurality of headsets used on the platform, which happened back in 2018

            TBH I don’t really care if the change was some nefarious plot to deliberately distort the data in their favour or just an “unforeseen consequence” of an entirely innocent but statistically illiterate attempt to provide more reliable data (although to me Occam’s Razor would tend to suggest the former, given that I doubt there are many dummies working at Valve). I’m sure it’s well within their technical capability to reliably detect the presence of any VR headset connected to the PC at any point when Steam is running whether or not it happens to be used to play a game which specifically works only via Steam VR, but they apparently chose not to do so.

            All that really matters is that the actual result of their changes was to dramatically increase the proportion of Index’s reported and dramatically reduce the relative proportion of Oculus headsets reported. Anything else is speculation.

            As to your point that they could have done this at any time. Well, sure, I just don’t think they cared that much before. They were perfectly happy with the survey under-reporting all VR users in the past (if they didn’t happen to have their headset plugged in at the time of the survey and sometimes even when they did), but after the previous data showed that Oculus headsets were now in first and second place (even amongst users of Valve’s own store) and that the Rift S was outselling the Index by a huge margin they perhaps felt it was time for a tweak, coincidentally just in time to be in place ready for the post Half Life: Alyx launch data.

  • Epxzz

    Thanks for the interesting article. I have a few questions… Monthly connected Headsets on Steam is actually a linear trend. It is independent of the underlaying data you use to calculate the absolute number. If this underlaying data is exponential, the resulting trend for VR Headsets will be exponential. You do not mention what data you use to demystify the % amount. The Number of monthly active users is not exponential, the number of concurrent users is not exponential. I did not find numbers for total registered steam accounts, maybe this is exponential (but it will count every double-registration and spambots). I would be interessted in the source or the formular you used. There are also a few month missing in the data set (Aug 2017-May 2018), why is that?