What makes a popular mobile game?
1. Background
The mobile games industry is developing so fast, with companies spending vast amounts of money on the development and marketing of these games to an equally large market. According to the newzoo, mobile gaming (smartphone and tablet) is largest segment of gaming market in 2019, which is 68.5 billion dollars (45% of the global games market). Of this, 54.9 billion dollars will come from smartphone games.
In this article, I will try to find what makes a popular game. I used the data of 17007 strategy games on the Apple App Store. Is was collected and made by the kaggle user Tristan. I downloaded it from the kaggle datasets.
2. The parameter of popularity
What makes a popular strategy game? To answer this question, we need to define a parameter of popularity. The download count will be the best choice. But the data I have did not contain the download count. Instead, we can use the ‘User Rating Count’ in the dataset as the estimator of popularity.
However, the User Rating Count is not so comparable between games. Different games have differnet release date and life period. Games which released earlier had more time to accumulate more rating counts than newly-released games. So, It is not fair to compare the User Rating Count between old and new games.
To make User Rating Count comparable, I made a new variable called ‘User Rating Count per Day’. It is calculated as:
Days from Release = Cuerrent Date - Original Release Date (Cuerrent Date is 3rd August 2019)
User Rating Count per Day = User Rating Count / Days from Release
Plot the distribution of User Rating Count per Day:
The distribution is highly positively skewed: Most of the games have the User Rating Count per Day nearly 0. To know more about the distribution, check some statistics:
count 7228.000000
mean 3.477717
std 50.573804
min 0.001394
25% 0.010283
50% 0.040541
75% 0.256975
90% 2.000000
max 2972.500000
We can see that the max is 2972.5, but the 75% quantile is 0.2 and 90% quantile is only 2. This means that 90% of the game has less than 2 persons make reviews per day! We can know that the rule of moblie trategy game market is ‘winner takes all’: A few top ‘winner’ games attracted most of the users. Most of games are ‘losers’, whose user groups are small.
Next, I label the top 10% of games as ‘winner’ and the remaining 90% as ‘loser’. I will analyze the differnece between the ‘winner’ group and the ‘loser’ group and try to find what makes a popular strategy game.
3. Which kind of charging mode associated with popularity?
There are 4 ways for a mobile game to charge and make profits:
- Free download, with in-app purchase (Make profits on in-app purchase only).
- Free download, without in-app purchase (Probably make profits on advertising).
- Charge for download, with in-app purchase (Make profits on both paid download and in-app purchase).
- Charge for download, without in-app purchase (Make profits on paid download only).
Which one is the best mode to make a popular game? To answer this question, I need to check the difference of proportion of charging mode between winners and losers:
loser | winner | |
---|---|---|
Free download, with IAP | 0.528435 | 0.795014 |
Free download, without IAP | 0.299877 | 0.087258 |
Paid download, with IAP | 0.048417 | 0.066482 |
Paid download, without IAP | 0.123271 | 0.051247 |
We can see that the free download, with IAP mode is more common among winners compared with losers (80.0% v.s. 52.8%). Meanwhile, the free download, without IAP mode is less common among winners compared with losers (8.7% v.s. 30.0%).
We can infer that popular games tend to use free download to attract users and use in-app purchases to make profits.
4. Is average user rating associated with popularity?
Intuitively, the game with high rating tends to be popular, because only acclaimed games can attract and retain large amount of users.
In this part, I am going to investigate the relation between average user rating and popularity. First, check the average user rating of games among the winners and losers:
The winners’ mean user rating is slightly higher than that of the losers (4.4 vs 4.0).
Then, table and plot the difference of rating proportion:
Winner or Loser | loser | winner |
---|---|---|
Average User Rating | ||
1.0 | 0.002152 | 0.000000 |
1.5 | 0.009222 | 0.000000 |
2.0 | 0.023363 | 0.000000 |
2.5 | 0.046111 | 0.009695 |
3.0 | 0.075008 | 0.022161 |
3.5 | 0.132647 | 0.058172 |
4.0 | 0.237319 | 0.139889 |
4.5 | 0.348601 | 0.613573 |
5.0 | 0.125576 | 0.156510 |
We can see that the winner group have fewer low-rating games (rating less than 4.5, 23.0% v.s. 52.6%) and more high-rating games (rating more than or equal to 4.5, 77.0% v.s. 47.4%).
We can draw the conclusion that game with high rating (>= 4.5) is more likely to be popular.
5. Is size associated with popularity?
The larger games may have better and more content which can attract more users. In this part, I am going to investigate the relation between size and popularity.
First, check the mean and median of size among winners and losers:
Mean
loser 1.303953e+08
winner 2.950661e+08
Median
loser 68202496.0
winner 178507776.0
In terms of both mean and median, the winners have larger size than the losers. The median and mean size of winners is as twice of losers.
Next, I divided games into three groups based on size:
- small (less than 28MB, the 25% quantile)
- medium (between 28MB and 164MB, the 25% and 75% quantile)
- large (larger than 164MB, the 75% quantile)
Then, table and plot the difference of size proportion:
Winner or Loser | loser | winner |
---|---|---|
Size Stratum | ||
large | 0.219797 | 0.520776 |
medium | 0.511067 | 0.401662 |
small | 0.269136 | 0.077562 |
We can see that the winner group have much more large games (52.1% v.s. 22.0%) and much less small games (7.8% v.s. 27.0%).
So, we can draw the conclusion that large games are more likely to be popular.
6. Conclusion and Discussion
In this article, I analyzed the Apple App Store Strategy Games data and tried to find what makes a popular strategy game. Firstly, I defined a parameter of popularity: User Rating Count per Day. Then I studied the association between popularity and charging mode, rating, size. I found that games with the free download and IAP (In-app purchase) charging mode, highly rated and larger size are more likely to be popular.
However, the conclusion is not so solid. Firstly, the User Rating Count is not a good parameter of popularity, because not all users want bother to grade a game, even they are super fans. I think the Download Count may be better. Secondly, the causal relationship between popularity and the factors is not clear. It is hard to say that those factors ‘make’ popular games: For example, larger games may have better and more game features which attarct more users and make games popular. However, the truth can also be that game developers will only keep updating popular games, their cash cows, and discard unpopular games. And it is updating makes popular games larger and larger. In other words, popularity can lead to increase of size. Therefore, to get the real causality, more efforts need to be done.