Hopefully this is okay to post here. I originally posted it elsewhere but I thought people here would be interested in it and wanted to contribute something 
.
A hot topic recently in the media and on gaming forums has been the cut that various storefronts take when selling games. This is usually focused on Steam, who are believed to take a cut of 30%, which is comparable to the take that Sony, Microsoft and Nintendo take on consoles, and Epic who have lowered their cut to 12% and pay for this partially by passing payment processing costs onto consumers in regions where those costs are high.
The Steam figure of course, is misleading as it does not include copies sold through keys, of which Valve take a 0% cut. Unlimited key generation has been a feature on Steam for many years now, and allows developers to sell a game wherever they want, including other stores on PC which support it, such as Itch.io, Humble, GMG, and many more. It also allows them to distribute games through direct distribution, or give out copies to backers which supported the game during crowdfunding campaigns.
But how misleading is it really? When this is bought up as a topic, people often reply saying "only a small number of keys are sold that way", "nobody uses key sellers", etc. If that were true, then Valve's cut would still be close to the 30% people report. If only we had a way of figuring out roughly how many people bought their games this way.
So that's what I'm trying to do. This post was inspired by a post from Lashman on Twitter which I will post below. While it is unfortunately impossible to determine exact how many copies of various games people own and where they bought it from. Steam does provide a method of separating reviews by copies purchased on Steam or elsewhere. Lashman however only sampled a few games and the variation was high enough that I didn't feel fully satisfied with the results.
Other storefronts on PC
Obviously a key piece of context here is that for keys to be distributed, there must be other stores to sell through. There are tens of legitimate key sellers on the platform, and a list of many of them can be found on www.isthereanydeal.com. These stores also take a cut, and that cut can vary significantly from as low as nearly nothing (which will be explained soon), to figures comparable to the 30% Valve would take. I will mention a couple here, but this list is far from exhaustive.
First of all, Humble. Humble provide two methods of key selling. The first is through the Humble store, which takes a 10% to charity, and a 15% cut to Humble, accounting for a total 25% cut. The other method of selling through Humble is by using the Humble Widget on the developers own website. If a developer chooses to take this approach, they will receive a 95% cut of the sales, with 5% going to Humble. The cut is viewable from the developer FAQ page here: Humble Store FAQ For Developers
Itch.io are the next I want to mention. They offer one of the best deals imaginable to developers, and are the only store I know of to offer a variable cut which is chosen by the developer themselves. Itch set a default cut to 10%, but the variable cut has no limits and a developer can set this as low as 0% if they wish, or as high as 100%. Many developers do not do this, however. Itch.io also provide a widget, just like Humble, and the itch.io widget takes the same cut as through the store. Information about the widget can be found The itch.io embeddable widget and the variable cut (which they call "open revenue sharing") can be found
I wanted to discuss GOG too, but exact numbers are more difficult to find. We do, however, have multiple developers who have said they take the same 30% cut that Steam takes. I wanted to mention them as they are one of the largest competitors to Steam we have, operate completely separately, and do not provide Steam keys as they are strictly DRM free only.
The Model
I use the same model used by Lashman, but extend the sample by using public data obtained through the steam API to find the number of user reviews for copies sold through Steam, or through some other approach, for the top 100 games over the past 2 weeks. The list of 100 games is generated by data provided by the SteamSpy API, and the code which generates this data was kindly provided by Wok who also developed the SteamSpyPi library which was also used here. The number of reviews are calculated directly from querying Steam. The code, which I have since modified slightly, is available from his Github: woctezuma/steam-cut
The model can be summarized by the following:
That summarizes everything hopefully. If I have forgotten something, let me know and I'll try and explain it.
The numbers
The numbers generated by the code are given in the following table, and a more human readable version provided in this Google spreadsheet:
The results show that on average, 72% of games are purchased through Steam, while 28% are purchased through third parties, including key resellers but also including retail copies. Depending on the game,this can go as low as just 9% of copies being sold through Steam and an effective Valve cut of 2.58% (a clear outlier). UPDATE: This outlier seems to have been caused by a bug, the real lowest should be 25% of copies sold on Steam, for an effective Valve cut of 6.17%. The "game" with the highest portion of copies sold through Steam in the top 100 was... Wallpaper Engine. I guess I didn't filter out software. Wallpaper Engine sells the vast majority of their copies directly through Steam, with an effective Valve cut of 27.98%. However, this should not be surprising since that particular piece of software does not seem to be available in many other places, if any.
For an average Valve cut, this data set approximates that Valve on average take a 20.07% split in revenue from games available on Steam, which is lower than many of their competing stores. However, as we previously said, Valve games do not make any sense being on this list, and once they are removed the average drops slightly to 19.78%. If anything is taken from this long post, that's probably the number to take. It is also worth noting that because of the new way Steam revenue sharing works, with popular games getting a better cut, this number likely overestimates the true value.
Final Notes
As I previously mentioned, the purpose of this was to calculate the amount Valve take from games distributed through Steam. This obviously has many limitations, like the fact that Valve have the financial burden of paying payment processing fees, or how the Steam payment cards they distribute in emerging markets are expensive for them to distribute (taking up to 15% of the value of the card). As a result, Valve themselves obviously do not get all the money from their part of any cut.
Another big point which has come up is how some people believe my numbers are misleading, as the 19.78% or 20.07% figures aren't what devs actually take home. My original intention was to focus on how much money Valve themselves take, and how once you account for key generation they actually take less than many of their competitors, and not what developers take home. I believed I addressed this by mentioning that other stores (where these steam keys are usually distributed) have cuts of their own, but it is worth giving it a second mention here apparently.
If we are to assume the highest a third party takes is 30% (which is true as far as we know) then the split that a developer takes back is 70%, compared to 30% to whatever stores they sold it on. That is the upper bound, a good lower bound is a little harder to calculate, and an accurate number is impossible as there is no way to calculate where each game was sold. The lowest cut that I can find for a third party is the 5% humble widget. However very few keys are likely to be sold this way. Instead, for my lower bound I am going to pick a 10% cut to the store, which is the default cut for Itch.io. If we assume all Steam copies sell at the 30% rate again (which as previously mentioned, is now an overestimate), and all third party sales are sold through Itch, here is what the above table would look like with these figures:
.A hot topic recently in the media and on gaming forums has been the cut that various storefronts take when selling games. This is usually focused on Steam, who are believed to take a cut of 30%, which is comparable to the take that Sony, Microsoft and Nintendo take on consoles, and Epic who have lowered their cut to 12% and pay for this partially by passing payment processing costs onto consumers in regions where those costs are high.
The Steam figure of course, is misleading as it does not include copies sold through keys, of which Valve take a 0% cut. Unlimited key generation has been a feature on Steam for many years now, and allows developers to sell a game wherever they want, including other stores on PC which support it, such as Itch.io, Humble, GMG, and many more. It also allows them to distribute games through direct distribution, or give out copies to backers which supported the game during crowdfunding campaigns.
But how misleading is it really? When this is bought up as a topic, people often reply saying "only a small number of keys are sold that way", "nobody uses key sellers", etc. If that were true, then Valve's cut would still be close to the 30% people report. If only we had a way of figuring out roughly how many people bought their games this way.
So that's what I'm trying to do. This post was inspired by a post from Lashman on Twitter which I will post below. While it is unfortunately impossible to determine exact how many copies of various games people own and where they bought it from. Steam does provide a method of separating reviews by copies purchased on Steam or elsewhere. Lashman however only sampled a few games and the variation was high enough that I didn't feel fully satisfied with the results.
Other storefronts on PC
Obviously a key piece of context here is that for keys to be distributed, there must be other stores to sell through. There are tens of legitimate key sellers on the platform, and a list of many of them can be found on www.isthereanydeal.com. These stores also take a cut, and that cut can vary significantly from as low as nearly nothing (which will be explained soon), to figures comparable to the 30% Valve would take. I will mention a couple here, but this list is far from exhaustive.
First of all, Humble. Humble provide two methods of key selling. The first is through the Humble store, which takes a 10% to charity, and a 15% cut to Humble, accounting for a total 25% cut. The other method of selling through Humble is by using the Humble Widget on the developers own website. If a developer chooses to take this approach, they will receive a 95% cut of the sales, with 5% going to Humble. The cut is viewable from the developer FAQ page here: Humble Store FAQ For Developers
Itch.io are the next I want to mention. They offer one of the best deals imaginable to developers, and are the only store I know of to offer a variable cut which is chosen by the developer themselves. Itch set a default cut to 10%, but the variable cut has no limits and a developer can set this as low as 0% if they wish, or as high as 100%. Many developers do not do this, however. Itch.io also provide a widget, just like Humble, and the itch.io widget takes the same cut as through the store. Information about the widget can be found The itch.io embeddable widget and the variable cut (which they call "open revenue sharing") can be found
I wanted to discuss GOG too, but exact numbers are more difficult to find. We do, however, have multiple developers who have said they take the same 30% cut that Steam takes. I wanted to mention them as they are one of the largest competitors to Steam we have, operate completely separately, and do not provide Steam keys as they are strictly DRM free only.
The Model
I use the same model used by Lashman, but extend the sample by using public data obtained through the steam API to find the number of user reviews for copies sold through Steam, or through some other approach, for the top 100 games over the past 2 weeks. The list of 100 games is generated by data provided by the SteamSpy API, and the code which generates this data was kindly provided by Wok who also developed the SteamSpyPi library which was also used here. The number of reviews are calculated directly from querying Steam. The code, which I have since modified slightly, is available from his Github: woctezuma/steam-cut
The model can be summarized by the following:
- All copies are assumed to operate at the old revenue share of 30% to Valve, 70% to the developer. As of December 1st this is no longer true and developers who make a game which generates over $10 million on Steam, will use an adjusted revenue share of 75%/25% on earnings beyond $10M. The revenue share is then adjusted again at $50 million, to a split of 80%/20%. Steam Community :: Group :: Steamworks Development
- Both groups of people (Steam purchasers vs Key purchasers) are assumed to review games at the same rate. Unfortunately I cannot think of a way of modelling this, short of doing a survey of users. This is one of the main assumptions the approach takes, and it is possible that it does not hold true. Hektor also mentioned this in the PC thread here, and suggested that maybe key purchasers are more likely to be engaged with the platform and the various features it has, so might be more likely to leave a review. I'm not going to get into that, and just leave it as a weakness of the model.
- Games are sampled from the top 100 games on Steam as of the player numbers in the past 2 weeks. This list includes many F2P games, like DOTA2, Warframe, etc, and those games are all excluded from the calculation. Valve games are kept in the data as they are sold through third party stores, but the "Valve cut" column should be ignored as obviously the developer cut of a Valve game would also go to Valve.
That summarizes everything hopefully. If I have forgotten something, let me know and I'll try and explain it.
The numbers
The numbers generated by the code are given in the following table, and a more human readable version provided in this Google spreadsheet:
Conclusions
The results show that on average, 72% of games are purchased through Steam, while 28% are purchased through third parties, including key resellers but also including retail copies. Depending on the game,
For an average Valve cut, this data set approximates that Valve on average take a 20.07% split in revenue from games available on Steam, which is lower than many of their competing stores. However, as we previously said, Valve games do not make any sense being on this list, and once they are removed the average drops slightly to 19.78%. If anything is taken from this long post, that's probably the number to take. It is also worth noting that because of the new way Steam revenue sharing works, with popular games getting a better cut, this number likely overestimates the true value.
Final Notes
As I previously mentioned, the purpose of this was to calculate the amount Valve take from games distributed through Steam. This obviously has many limitations, like the fact that Valve have the financial burden of paying payment processing fees, or how the Steam payment cards they distribute in emerging markets are expensive for them to distribute (taking up to 15% of the value of the card). As a result, Valve themselves obviously do not get all the money from their part of any cut.
Another big point which has come up is how some people believe my numbers are misleading, as the 19.78% or 20.07% figures aren't what devs actually take home. My original intention was to focus on how much money Valve themselves take, and how once you account for key generation they actually take less than many of their competitors, and not what developers take home. I believed I addressed this by mentioning that other stores (where these steam keys are usually distributed) have cuts of their own, but it is worth giving it a second mention here apparently.
If we are to assume the highest a third party takes is 30% (which is true as far as we know) then the split that a developer takes back is 70%, compared to 30% to whatever stores they sold it on. That is the upper bound, a good lower bound is a little harder to calculate, and an accurate number is impossible as there is no way to calculate where each game was sold. The lowest cut that I can find for a third party is the 5% humble widget. However very few keys are likely to be sold this way. Instead, for my lower bound I am going to pick a 10% cut to the store, which is the default cut for Itch.io. If we assume all Steam copies sell at the 30% rate again (which as previously mentioned, is now an overestimate), and all third party sales are sold through Itch, here is what the above table would look like with these figures:
As you can see, the share is between 20-30% across the board with a few exceptions dipping into the 18-19% range, with the remaining 80-70% to the dev. If you want to maximize the share to developers, buy through widgets or through Itch.io who offer the best share, and try and argue for a reduced cut across the board including from console manufacturers. However, there are other things which make this more complicated, like how developers get extra visibility from the metric boost that copies sold through Steam give them. See Paz's posts in this thread if you want to read more about that: Would indie developers rather I bought their games on Steam or Itch.io?.
Last edited:
