Big data behind Steam: game developers end up with only 38%

Not long ago, Valve just announced the 2019 list of Steam platforms. Although many games are divided into four ranges of platinum, gold, silver and bronze according to income, but due to opaque data, it is often difficult to judge the income range of a Steam game .

Recently, two independent game developers used their own methods to dig deep into the data of thousands of Steam games and gave their own judgments:

Behind the revenue data: only 38% of developers have it

As competition in the gaming market intensifies, how to measure the return on investment has become a concern for all peers. For an independent game developer, if they can have a certain understanding of the game income profile, the risk assessment can be made more accurate. But just how much revenue do we see independent developers behind Steam sales?

Independent developer Danny Weinbaum gives an actual income calculation method:

Developer income = number of owners * USD price * 0.93 (VAT) * 0.92 (refund rate) * 0.8 (average regional pricing) * 0.8 (average discount) * 0.7 (platform commission).

Except for the dollar price and the platform fee, all data are approximate values. All the above factors are combined to get a multiplier of 0.38. In other words, if the game is priced at $ 1, developers can get one for each sale $ 0.38.

For an already released game,The first week’s income accounted for 13% of the total five-year income. In addition, three-month income accounted for 33%, one-year income contributed 58%, two-year income accounted for 75%, three-year income accounted for 87%, and four-year income accounted for 95%. %.

So, if a game is released for 7 months, then in this way, the income of seven months is almost 40% of the total income of five years. With the increase in the number of games, and the slow increase in the number of Steam users and users, the game success rate has decreased year by year.

However, another developer, Sergio Garces, said that Danny’s missing data restricts the age of the game and that Early Access’s end statistics are inaccurate. In addition, some prices are wrong due to statistical problems. To this end, he used the Steam API excuse to capture the data of the last week of 2019. The statistics covered the game with at least 10 reviews and 1000 owners, and the total number of games was 7,000. The following conclusions were reached:

65 is the best “guessing” number (that is, the number of reviews multiplied by 65 is the predicted revenue); many game owners do not write reviews, and we choose to ignore these situations when predicting revenue, because the larger the multiplier, the more revenue The higher the result; 30-100 is a better range because it covers most games, and 65 is the average. This range covers most low-selling games. Of course, mistakes also underestimate revenue.

After counting the number of owners, we need to convert it into income. Obviously, this is not as simple as multiplying the price of US dollars. Sergio thinks that Danny’s income calculation method is very good, and it is also a very useful way to predict long-term income. Because it can predict future revenue results based on current game sales. The prediction method in this article mainly looks at the amount of reviews at different times. For this reason, more than 1,000 reviews in the first five years of the game are called. After visualization, the following chart is shown:

This graph covers the data of more than 400 games. The red line is the daily average, the blue is the median, and the green line is used by Danny in his article. From the trend point of view, most of the income is concentrated. In the first year, there are more than you think, so this approach is relatively conservative.

Sergio Garces used historical evaluation data in his analysis to create simpler progress models to predict future evaluation numbers. He said that this is not complicated, “because the revenue curve of some games is linear, while others are logarithmic. There are both ways Adopt them and use them to generate a range based on “best guess” numbers to get the most realistic results possible. “

What needs to be added is that it is difficult for us to predict what will happen in five years time, because we don’t have enough data yet, and the game market will change a lot, so the current rules do not apply to the future situation.

Growing Steam platform: more high-income games

As we all know, the number of games on the Steam platform has been exploding in recent years. However, the number of new game releases has reached its peak, and we even see a downward trend in the number of new independent games in 2019.

Some people say that it is becoming more and more difficult to make independent games. This is actually supported by data. In the games released in recent years, the median has been declining. We can see from the following data:

But Sergio wants to look at the problem from another angle. Instead of looking at the percentage of game success rates, we compare it with absolute numbers. After all, more and more people are publishing games, so the number of winners also increases:

This is indeed true. Even as more and more games fail, it is also true that more and more people have succeeded, which is encouraging.

Scoring matters

Danny used the following diagram to show the comparison between the score and the income, but Sergio remade it:

The median is used instead of the average, so there is no need to calculate income forecasts. Interestingly, Sergio’s data did not fall by 70% as Danny showed, so the “indie game doom” said that it is more confusing audio and visual noise.

Obviously, if you consider player purchase behavior, it is normal for high scores to bring high income. Many players read reviews before buying a game. The lower the rating, the lower their willingness to buy. Conversely, if the game is of high quality and liked by players, they will be more willing to recommend it to their friends, and word of mouth is of great importance to the success of Steam games.


The market as a whole can understand some trends, but in order to get more information, we need to split it, the best way is to use the Steam tag.

Chris Zukowski has written a good article mentioning that he used tag data for marketing before making a game. Understanding a field and using data to analyze it is the right approach, rather than just looking at the data and extracting some labels to analyze the reason for the success of the game.

The most used tags on the Steam platform:

Screenshot of Steam platform game (part) forecasted revenue data:

Source: GameLook


Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *