As an independent game practitioner, I have been thinking about the market survival of different types of games. My hobby sounds a little weird: browsing the Steam store, trying to analyze the game, estimating sales, and figuring out why the game was successful or not.
Until one day, I had a flash of light: Is it possible to make a steam game list, the information in it includes the game tag, genre, release date and approximate income? I find it interesting to find some patterns from these data (especially labels). So I started collecting data.
Note: The title of this article says “type”, but on a technical level, this article studies labels, which are more subdivided than types, and I think it is more useful. In the text, I will use both of these words, because in my opinion labels are types in a narrow sense.
Estimate game revenue
This is the simplest estimation method for independent game developers. First estimate how many people buy a game: multiply the number of reviews by 50 (according to Jake Birkett’s research, the range is 30 to 100, and 50 is a compromise number). With the number of purchases, multiply it by the US Pricing is multiplied by 0.38 to estimate net income.
Why multiply by 0.38:
Number of purchasers * US region pricing * 0.93 (VAT) * 0.92 (refund) * 0.8 (average regional spread) * 0.8 (average discount) * 0.7 (platform commission)
This is, of course, an estimation, because the actual net income of a game will vary greatly depending on the discount rate and the sales area (affecting the average regional price and the proportion of VAT paid). In addition, as the game’s discounts increase during the cycle, the average discount multiplier will gradually decrease. 0.8 is the average discount of our game (Eastshade) 8 months after its release. The biggest discount so far is 40%.
Another important factor in estimating the final revenue of a game is the position of the game in its sales curve. The recently released game is just the first phase of revenue. To infer lifetime income, here are some useful data:
On average, the percentage of revenue (cumulative) for a game over its lifetime is as follows:
- 13% after one week
- 33% after 3 months
- 58% after one year
- 75% after two years
- 87% after three years
- 95% after four years
- For convenience, let’s assume that the game’s lifetime income is 100% 5 years after its release (at this time the sales volume will be very low)
These data, like many of the data in this article, are estimates. Some games have a larger long tail, earning well over 5 years, while others sell for less than 5 years. I determined this data based on sales data collected from other games, including our own, as well as reviewing charts of various games on steam, and watching the trend of decreasing reviews over time. As long as you know the current approximate revenue of a game, you can use this data to predict its lifecycle revenue (we can use the number of reviews mentioned earlier) and the game release date.
Current estimated revenue * (1 /% of current life cycle)
Assuming that a game has been released for 7 months, you can use our chart to see that statistically, it may have reached about 40% of the final revenue, so we can put this number into the “current life cycle percentage” In, get a multiplier of 2.5.
Note: If a developer’s project has been released for many years, please tell me if my estimated sales curve matches your actual data. Thank you!
Thanks to Jake Birkett’s article, I’m very confident in the accuracy of the data in the first year of sale, but the curve will become more difficult to collect data later, so I’m not quite sure.
This is how I estimate and predict revenue in my data. It must be noted that this method is simple and rude and prone to some problems. For example, if the current price of the game is lower than when it was originally released, the result of my formula will be lower. Another big problem is that the average discount rate varies widely, and I don’t have data on discount history to correct this. Of course, each number in the formula has a certain variable range, so the difference between the estimated result and the actual number is 50% to 300%.
Note: If developers are willing to share with me the gap between my formula and your actual data, it would be great.
This table covers almost all games on Steam.
Note: The EA stage game is not good because the script extracts the date of the 1.0 release. Then those games that I have age restrictions can’t pull the data.
Making games harder and harder to make money?
The median and average income of steam games is declining, which is often used to quote the so-called “indiepocalypse”. But from another perspective: the percentage of games that earn considerable revenue. As we all know, there are a lot of games on steam now that are crudely made or do not pursue commercial success. I personally think that steam’s algorithm will bury games with low conversion rates. So, for example, if 30% of Steam games are expected to have a net income of more than 200,000, as long as your game is in the top 30% in terms of product quality and marketing, there is reason to expect you to have more than 200,000 net income opportunities. Let ’s take a look at these data: ‘
The bad news is, yes, the percentage has been declining. But it seems to have stopped. But while the average income has fallen, the number of games has increased. How many games are our opponents to watch out for? It turns out that many of these games have fewer than 10 reviews. Take the liberty to say that most of these games do not compete with full-time independent developers. As you can see below, the situation looks a little optimistic when we filter out games with less than 10 reviews. This year (so far) the odds of success and median income have even risen nicely.
So is the game industry getting harder? This is hard to say, since Steam ’s average revenue per game has been declining in most cases since 2013, but developing games is easier than ever. In fact, putting games on Steam is also easier than ever. Easy at any time. The influx of low-cost games reduces average revenue, but is the high-end gaming market actually more competitive? For a hard-working game developer, it is still difficult to make money by creating quality products and operating it seriously. My gut tells me, yes, it’s getting harder, but the situation is not as severe as the declining average income. Just take a look at the games that are launched, and check if you can make a quality that is comparable to yours. That’s why you should choose the type of game that has market demand. Different tags on steam have different odds of success. We will discuss this in the next section.
Looking at revenue from game tags
The idea actually popped up when I browsed the tags on SteamSpy. I noticed that the median income of the labels did not match my intuitive perception of certain types. Some tags have very high median income, but in my opinion they are low-demand types, while other tags have very low median income, but I think they should be relatively popular. I suddenly thought that the median doesn’t tell the whole story. For example, if one type is particularly prosperous or depressed, say 40% makes a lot of money, and the other 60% fails-earning less than $ 5,000, the median value will look very low. But I think this type of game is very feasible for a full-time independent developer, because a hard-working developer will not compete with the bottom 60% of people (I want to say that a hard-working developer is actually It’s just competing with the top 10% of games on Steam, and the rest are not.) The average is even less referenced, because a few big works will cause large fluctuations in the data. I think this statistic is a good reference: the percentage of games that earn more than X dollars-because it shows exactly the benchmark of the game you have to compete with. Alright, now the fun part!
Tag Data All Time
Tag Data Since 2017
You can sort the rows by right-clicking the letter of the column and clicking “organize a-z / z-a” (this will only change how the worksheet is looked up). In the Revenue by Tag table, the tags for each income range are linked to a collection of games, so you can click to see what games are included.
Obviously, the most interesting part of this article is the data itself. You can now read the data carefully and draw your own conclusions. Here are my observations:
I ignore low count (30) tags because there is not enough data. I ignore the high count (2000) tags because they are too vague to be a useful classification (“indie” is the most extreme example). This data taught me how useless the “independent” label is at this point. I recommend against using it to tag your game at all.
Some label statistics are clearly misleading. Batman is statistically the best label ever, but there are very few games (barely meets my minimum standard, 10, otherwise it will not be listed at all), and all games have IP authorization, this label only applies For companies with adequate budgets. There are many such tags.
Based on these data, the market viability of the “cat” label is moderate. It seems that adding cats to the game alone does not lead to good sales. There are many mini games in tags, but the same is true of many better-performing tags.
Moddable tags perform well. I removed this at the beginning, and I thought that only games with large budgets would have enough size and community to support the mod suite. But after some research, this may not be the case. There are more than 500 games, and there are many small and medium cost games. You can pay attention to this label.
Since my script cannot extract age-restricted games, some tag data, such as Horror, may be inaccurate.
Does the evaluation score matter?
Once we have the data, we will find such a problem. The following is a comparison chart of evaluation scores and income. I grouped the games by percentage points (a total of 100 groups) and then calculated the average expected revenue for that group of games. I excluded groups with less than 50 games, because at this low order of magnitude, a high-selling game with a poor rating may break the average score (such as a poorly rated 3A). That’s why the chart starts at 30. Only a few games have scores below 30, so the data is very noisy. I also excluded games that made more than 5M to prevent outliers from making the data more noisy.
I used to believe that scoring was not important. I have seen many articles say this, and I have heard some publishers say the same. And this chart is a complete counter-example, at least in general general trends. The correlation between ratings and income is very clear, even if the data is a bit noisy. I’m not sure what’s happening around the 70% area.
to sum up
I’m by no means the first to say this: I think the most important marketing decision for a game developer is to choose the type of game. I believe the current market demand for game genres is critical to the possibility of games recovering costs. Of course, there are some games that reveal untapped markets. Indeed, those groundbreaking games have risen above the usual market size for their genre. But in general, the sales volume of games is comparable to the market size of game types. Therefore, when choosing the type of game you want to make, it’s wise to consider market viability (if you make a living from independent development).
The Steam tag is too simple and inconsistent and does not tell the whole picture. As far as the label revenue data is concerned, although I find it useful in assessing the market feasibility, it is interesting to me mainly by looking at it. To clarify, I collected this data mainly because I was obsessed with it, not as an absolute rule to decide the next game. I think case studies are more important than overall market research, and I also think that frank and intuitive assessments of individual games are more useful than big data.
Author: Danny Weinbaum
Source: Indietavern Translation
English original text from Gamasutra
Original address: genre_viability_on_steam_and_other_trends__an_analysis_using_review_count