Steam Hardware and Software Survey (May 2017)

[CarstenS]

I am not sure if this is a regional thing because of privacy laws, so: Is it for you guys in other regions than germany the case as well, that steam (has to?) explicitly asks if you want to take part in the survey and which data is going to be sent to Valve?

Just thinking whether or not sample size could be materially different from install base.

 

[entity279]

They (steam) ought to make sure that who they're sampling (surveying) accurately represents the whole install base. If people who reject answering form a distinct cathegory otherwise unrepresented, then indeed the survey is flawed by definition (so why have it if true? ).

However this problem is not new, and thus i expect it to have an accepted solution. For instance, pre-election polls. In my country they survey people via phone. I suspect they used to call landlines only (i've been surveyed 3-4 times, always by a call to my landline number) What about people that don't answer? That don't have a phone? That don't own a landline ( amost nobody before their mid late 30s does)?

 

[BRiT]

In the US, Steam would ask for your acceptance to submit the survey.

With that said, the very last time Steam asked for me to submit data was back around 2010. I have never been prompted to submit hardware survey data since then regardless of how many games I played or purchased or added.

 

[Silent_Buddha]

As of 2015 Steam has over 125 million registered users (https://www.vg247.com/2015/02/24/steam-has-over-125-million-active-users-8-9m-concurrent-peak/ ). It's likely even more now.

At the time I made this post there were over 10 million Steam users logged in. A few hours ago there were 13 million users logged in. (http://store.steampowered.com/stats/ )

If Steam surveyed 1.2 million users a year (100k users a month a relatively massive sample size), a person could easily go 10-20 years without getting a survey. Hell with how many people are on Steam, it's entirely possible that a person will never get surveyed. This isn't to say that Steam are surveying that many people, just pointing out that it's highly unlikely that every Steam user will get surveyed in any given year in any 10-20 year time frame much less get surveyed twice.

To put that into perspective, a Political survey usually surveys less than 1000 people. (http://edition.cnn.com/ALLPOLITICS/1997/gen/indepth/polls/faq.html )

How Many People Are Interviewed?
It varies from poll to poll, but usually Gallup and Yankelovich will interview 600 to 1000 people. In almost every case, those people (pollsters call them "respondents") will be at least 18 years old.

Obviously there's an error margin with any sort of poll or survey. I think it's safe to say that the Steam Survey is likely pretty accurate with regards to users of Steam. Now, you could argue that Steam isn't representative of the gaming market, but even then it'd at least be close as Steam is by far the largest gaming portal for PC users.

Regards,
SB

 

[swaaye]

I run Steam on a quite a few different computers. I would say I get survey pop ups 5 times each year across these various machines.

 

[Silent_Buddha]

It's random, as it should be. I've only ever gotten it once and I've been using it since HL2 game out.

Regards,
SB

 

[BRiT]

It would be better if it was everyone at least once a year.

 

[Geeforcer]

So the last two months the results have been... very atypical. And I am going blame PUBG.

First of all, there seems to be dramatic expansion of Steam-acessable PCs, as can be observed by 1080Ti numbers. Up until September it was following the typical “highest-end” uptake trajectory seemingly in the way to .85-1.15% percent where the cards at this price point settle at. Instead it stalled in September and collapsed in Oct. Since no one in their right mind is “upgrading” from 1080Ti (or plain 1080, for that matter) the only explanation that makes sense is a very significant expansion of survey participants.... which would also explain a unexpected mega surge of low-end/obsolete Nvidia hardware.

So what caused this Steam surge? I am going to blame PUBG popularity in Asia. Users (individuals and cafes) are installing Steam client on their machines for the first time OR going out and buying inexpensive PCs to play it, which more likely then not will be whatever the cheapest non-crypto mining cards are. Either way, the survey is suddenly hitting a whole crapload of low-end PCs for the first time (up to 50% more if we compare where 1080 cards should be Vs where they are now).

 

[DavidGraham]

NVIDIA share grew to 80% of the survey now because of this sudden surge. I agree with you it definitely because of PUBG. It also exposed how popular the 960, 750Ti and 1060 are.

 

[Geeforcer]

This also exposes the dark side of crypto mining: I would be very surprised if AMDs market share by revenue did not hold or even increase for late Q3 and Q4 whilist the relative gaming installment base shrunk significantly.

 

[HMBR]

Simplified Chinese
56.37%
+26.83%

pretty amazing, is that because of PUBG?

also
Windows 7 64 bit
63.60%
+22.59%

China is not a Windows 10 fan?

 

[Grall]

Ugh, win7 is seriously outdated for gaming these days. You don't even get DX11 on 7, do you?

Maybe vulkan works under win7. (All 3 games or whatever that supports it...)

 

[Geeforcer]

D3D 11 was back-ported all the way back to Vista. Win 10 was the D3D 12 transition point.

 

[Geeforcer]

Simplified Chinese
56.37%
+26.83%

pretty amazing, is that because of PUBG?

also
Windows 7 64 bit
63.60%
+22.59%

China is not a Windows 10 fan?

Lol, I was so focused on trying to deduce things from the GPU data that I missed that huge neon sign. I now really wish I could see the trend in language breakdown over the last couple of months to help confirm my ~50% user base expansion theory.

 

[HMBR]

the first D3D11 cards and game reached the market in late 2009, when Windows 7 was fairly new; Vulkan also works fine on Windows 7, Doom with Vulkan on Windows 7 and 10 seems to perform the same.

hmmm, was steam blocked in China until recently or something?

 

[Geeforcer]

 

[Kaotik]

The survey is still worthless, since Valve still won't tell how it's measured and won't correct clear mistakes (like single HD 8800 model, only sold by OEMs machines for one year, being more popular than whole HD 7800 -range based on same chip and sold for several years)

 

[DavidGraham]

New The survey is still worthless, since Valve still won't tell how it's measured and won't correct clear mistakes (like single HD 8800 model, only sold by OEMs machines for one year, being more popular than whole HD 7800 -range based on same chip and sold for several years)

In light of the recent developments, It's obvious now why that's happening, a sudden influx of OEM machines was greater than the percentage of leftovers from HD7800 generation after consumers have migrated from it. There are no data mistakes here, just misrepresentations due to various factors.

 

[Silent_Buddha]

pretty amazing, is that because of PUBG?

No, but it certainly helps. The biggest thing was Steam finally getting official support and approval for use in China from both Valve and the Chinese government. It's still a distant 2nd to Tencent PC distribution, but has grown significantly since their official launch in China.

I believe H1Z1's King of the Hill (their version of battle royale) is still massively more popular in China than PUBG. But that could be slowly changing.

Regards,
SB

 

[Kaotik]

In light of the recent developments, It's obvious now why that's happening, a sudden influx of OEM machines was greater than the percentage of leftovers from HD7800 generation after consumers have migrated from it. There are no data mistakes here, just misrepresentations due to various factors.

It's been like that for years, even with sudden influx of OEM machines the HD 7800 -series started selling retail a year earlier and was still sold after HD 8800, there's simply no way for those numbers to be right.