Numerical characteristics of online games

    8 years ago, a small development team, almost former students, decided to make their first big, as they thought then, game. They performed all the functions mixed up. And game design, and programming, and art. Who knew what. The resulting game was called Retribution. It was a browser. She showed good results: income from the game grew from several tens to hundreds of thousands of rubles in the first months. For a group of several people this is already some kind of money. In this mode, the game lasted about a year. And then the audience began to decrease, income began to fall. The developers thought: "Probably not enough content." The game was content-dependent, i.e. constantly lived on updates. They pushed, began to produce more locations, items, levels, craft recipes - everything that they did before. But users still continued to leave. And it was not clear why this was happening. As a result, the project, having existed for three years, was closed - sooner or later, all online games are waiting for such an end.

    The art of the character of the browser game Retribution

    Then we did not realize how important analytics and understanding of the processes that go on inside the game. Intuitively, we did everything to make the game die faster, although it was not so difficult to correct the situation and give the project another year. Now we already understand this. I would like to talk about the importance of analytics in online projects here.

    What is analytics for?

    So, first of all, analytics is necessary to understand the essence of the current situation that has developed inside an online game. The same can be said of any other online project, for example, an online store, forum or entertainment portal. Analytics suggests the developer to look at the gaming community and at the game as a whole as a complex mathematical system, the dynamics of which are described in some approximation by fairly simple models. Secondly, when we understand what is happening in the game, we set ourselves the goal - to influence the ongoing processes. There is a need to understand what indicators you need to have and how to reach the calculated values.


    Let's start with conversions. The term "conversion" means any action that we want to receive from a player, or rather, the proportion of players who performed the desired action regarding the total number of people who could perform it. For example, the most commonplace is the conversion to registration. So that a person visits the site or application, and then signs up. Accordingly, we consider the percentage of registrants from all visitors. Say, 10% of those registered are already a good value. In some games it is much higher, in some lower. Conversions are very dependent on the platform on which the game is oriented. In mobile games, some conversions, in browser and social networks, are different.

    In browser games, the total conversion may be higher than in those where the user must download the client part. However, it happens and vice versa. For example, many European gaming portals, such as Kongregate, recommend that their users opt out of Google Chrome. It would seem a very popular browser. And the reason is very simple: Chrome has stopped supporting Java Runtime - a plugin used by individual games. As a result, games running on these portals are blocked. Therefore, they recommend switching to Internet Explorer 11 or Firefox. This happens very often, a serious problem. For example, a few years ago we collected statistics and found that almost no user installed the Unity plug-in for the browser. Now the situation has changed. But Unity is still a long way from Flash Player. Flash coverage is very good: most users have a Flash plugin. However, we are aware of a number of other problems associated with its support, especially on mobile platforms. On such things, projects lose a significant part of the audience.

    When we talk about conversions, we need to understand that the exclusion of one type of conversion does not automatically increase the total conversion by its value. For example, half of people fall off on registrations. Let's do the so-called pending registration. So that a person does not need to enter a login, password and so on. He clicks “Play”, he is assigned an ID, and that’s it. It would seem that everyone should immediately get into the game, and the total conversion should increase. In fact, practice shows that users who did not fall off during registration will fall off later. Therefore, the total conversion will increase not by 50%, but much lower.

    Art from the Panzar game download page

    When I worked in the Panzar studio, we conducted a study of the audience of the client game of the same name. We made a delayed registration procedure for her when the user downloads a special launcher program. It weighed about 10 MB and automatically started downloading the rest of the content, a person did not need to do anything. He could register directly in launcher or run the client without registration, and registering on the site was optional. It seemed to us that this would increase conversions, more people would install the game and launch it. But everything turned out the other way around - the situation worsened. For ourselves, we explained this by a decrease in motivation. When a person gets something for free, and he does not exert any efforts even to register, then it is much easier to part with it. Many users did not even download the client part. When we returned the usual, classic registration, the percentage of those who finished the game increased significantly. We set up experiments with delayed registration on different traffic channels in order to eliminate the induced effect.

    Another conversion that needs to be mentioned is the conversion to a paying audience. This is one of the very important values ​​that indicates how much active audience regularly makes payments. The conversion to a paying audience usually does not take into account the so-called random payers. They are classified according to the frequency of payments and the size of the average check. The gaming community has a terminology comparing the payer with the inhabitants of the ocean. So "whales" are called those who make the maximum payments, often multi-million. Such an audience is very small, not more than 5% of all paying players. But the revenue from them can be significant and depends on the specific game. Usually, whales make a significant contribution to the income of projects with the Asian model of monetization.

    Audience size

    Speaking of conversions, one should not forget about the so-called “effective registration” or “effective loading”. This most often means a user who logged into the game a second time during a certain period. Such a person is considered already an effective registration or, in other words, an active player. The criteria are platform dependent. For example, in the mobile segment, if a person just started the game a second time, then most often he is already considered active. We are interested in measuring primarily the volume of the active audience, and not the total number of registrations or downloads in the project. When we just talk about traffic that came to our site or at the entrance gate of the game, this does not give us any important information that can be used. Therefore, criteria such as DAU, WAU, MAU, etc., which describe the volume of an active audience, are often introduced.

    It is clear that in client games we can operate with monthly periods. Mobile games use shorter periods of time. But these are not the only characteristics describing the scope of our project. There are also such important quantities as CCU and PCCU - the number of players online. Peak CCU - the maximum amount for a certain period of time, say, per day. For different analytical tasks it is advisable to use different characteristics, it all depends on the situation, on the calculations that we want to conduct and what values ​​we want to calculate. There is a definite relationship between MAU, DAU, and CCU. Thanks to it, you can approximately derive one characteristic from another.

    Audience loyalty

    In addition to the volume of the audience, we operate with such a group of statistical characteristics as “loyalty” - involving the audience in the game process. There are different criteria for evaluating a project. Previously, most often they talked about such a value as retention, that is, about the game’s ability to accumulate an audience. It can be calculated in different ways. For example, to take the difference between the number of monthly audience at the beginning and end of the period, subtract all traffic that came from advertising, attribute the result to the number of players who were at the beginning of the month, we get retention. If the retention of the project is 60%, this means that 60% of the audience available at the beginning of the period reached the end of the period. This is the simplest definition. There are others.

    Ret = (E - N) / S * 100%

    Here E is the number of players at the end of the period, N is the number of new players acquired during the period, S is the number of players at the beginning of the period.

    Recently, however, analysts have been more inclined towards the importance of a different quantity that describes the loyalty of the audience. This is the so-called sticky-factor or “stickiness”. In fact, it describes how much people like the game. For client, browser and some other games, it is considered to be considered as the relation of DAU to MAU. It correlates with the number of days in the period during which the player enters the game relative to the entire period. For example, if all players enter the game every day of the month, then DAU = MAU, this is 100% stickiness, perfect condition. But this does not happen. Usually cool enough when the sticky-factor is at 20%. Let's say Mail.Ru mini-games have extremely high stickiness for such a casual project. For many games of the company, this value varies in the region of 16-20%. Some hardcore client games, such as Allods Online, also have quite high rates of retention and stickiness. No less high indicators can generate social networks, a significant proportion of users of which visit such projects daily.

    So, the sticky factor is calculated as the ratio of DAU to MAU and characterizes the involvement of the audience in the project, and retention characterizes the ability of the project to accumulate the audience. When one speaks of “devotion”, other quantities are sometimes used. One of them churn rate - burnout or loss of audience. This, in a sense, is the reverse retention - a value proportional to the departure of the audience.

    You must understand that there is a natural obsolescence of the project - a period when all indicators begin to fall. This is an almost inevitable process. There are units of games on the market that can work for decades - for example, World of Warcraft.


    However, counting conversions, the size of the audience and its dedication is not enough. An important role in the group of statistical variables is played by the characteristics that describe the virality of the project. For example, the so-called coefficient of attraction (virality). He describes the so-called word of mouth. It is believed that the Angry Birds project successfully gained its audience precisely because of this. At the start, when the game was just starting, the project could not quickly gain popularity and audience, despite good marketing budgets. Only huge budgets and pre-installed games on a number of the latest mobile devices have a viral effect. Many analysts agree that the “birds” so abruptly gained an audience, having formed in the mass consciousness the understanding that Angry Birds is a brand, something popular and everyday.

    Let's take a closer look at the coefficient of attraction. When we inject advertising traffic into the game, its influence is not limited to the audience that came from the advertisement. For us, the delayed effect is no less important: the audience that will come later thanks to this advertising traffic. In this case, virality will work. For example, in the Panzar multiplayer session action, 50% of the audience that came to us from advertising traffic accounted for about 45% of the audience that came from virality. The remaining 5% was generated by the referral system. The figures are approximate, as as the project developed and the sources of traffic changed, they changed significantly, both in the direction of increasing the share of free traffic, and vice versa. In a rough approximation, traffic generated by the engagement rate,

    World of Warcraft Fan Art

    What kind of strange mythical viral audience is this? For example, when a person looked at an advertising banner, but did not click on it. Then he drove in the search engine the name of the game, because he saw the advertisement more than once and it was deposited in his memory, and finally wanted to come and see. Either a friend playing Panzar told the man, and he stopped by to inquire. Thus, we can say that advertising has a double effect. It leads not only direct traffic, but also creates a kind of delayed “tail”, in which there will be viral traffic. Viral traffic is good because it has very high conversions. If a person entered the search engine himself, entered the name of the game, found it, read about it, or switched from some kind of review, then most likely he is more interested than the player,

    By the way, such banners do not work very well. But a good result was given by some mailings with full-breasted naked elves, who in some places were covered with boxes of gifts, and a large button "Take" in the pelvis. On such mailings about 6% of the audience returned. And here we come to another parameter - the return coefficient. It also characterizes virality. A returning audience often costs us a lot less than attracting from scratch. The same World of Warcraft sifted through itself so much traffic that it became very expensive to attract new players. How can you estimate the value of one player in WoW? The most rough estimates can be obtained from referral. They have an invite a friend system. She gives a number of in-game bonuses. But the main thing is that if the friend you invite to the game, pay a monthly subscription, you also get a monthly subscription. Accordingly, we can conclude that the cost of one player in WoW should be higher than the monthly subscription fee, since such a system is beneficial for the company. It is believed that 1.5 - 2 times. Since the retention in WoW is quite high, and even if a player signs up, he pays immediately for several months, and paying 400 rubles for it is relatively cheap for Blizzard.

    In conclusion of the bloc on virality, let me remind you of completely captain's truth. Viral is translated from English as "viral". So, the success of many viral mechanisms lies in the viral effect on the audience, suggesting that players will share information with each other.

    Screenshot of the viral video "WoW Veterans"


    Key monetization options are ARPU and ARPPU. Average Revenue Per User, i.e. the average income from any active player is like the average temperature in a hospital. ARPPU (Average Revenue Per Paying User) is the average income from the payer. Again, by “payer” we mean someone who pays at a certain frequency. If a user made a payment only once in his life, then in these statistics we usually do not take it into account. It is reasonable to assume that the simplest estimate of project profitability can be obtained by multiplying MAU by ARPU. This is a rather crude but effective assessment.

    Speaking of monetization, we also use the term “time before the first payment”, which is relevant in browser and client games. In mobile games, in this regard, everything is bad enough. If a person did not pay in the first game sessions or on the first day, then most likely he will never pay. The sad truth of mobile games is that most players leave the project right away. Monthly retention of games on a mobile platform rarely exceeds 10-15%. In client games, this time can be up to a week, but it is very dependent on the game. In browser games - from 3 to 7 days, sometimes more.

    In addition to the described parameters, we also calculate the “volume of the shopping basket”. You can compare it with ARPPU and figure it out, isn’t it possible that a person does not pay us extra? And finally, there is such a thing as “price elasticity of demand”. It means how much the demand for a particular paid service will change with an increase or decrease in the price of it by 1%. It is important to understand that in games the elasticity of demand is significantly different from the real market. If the store doubles the price of a jacket, then we will lose a fairly large portion of customers. If in games you double the cost of a service, then we can well get a total increase in profitability of 20-30%. Even though part of the audience will stop buying. You should always experiment with this. For example, in Mail Mini-Games.

    Let's move on to the key parameters that describe the ability of the entire project to make a profit. One of them is LTV (Lifetime Value) - lifetime income from a player, that is, how much money one player will bring us, one registration, for the entire lifetime of the game. We must compare this value with CPA (Cost Per Action), the cost of attracting a person to the project. And if it turns out that LTV exceeds CPA, then perhaps the project will pay off. Otherwise, the project is unprofitable.

    Last but not least monetization characteristic in the list is ROI (Return On Investment). Sometimes they say ROR: Rate of Return. This is an indicator describing the total return on investment, that is, the return on the project taking into account the funds invested in it. Often we ask ourselves: how much money will our project bring if we invest a million in it? And two million? Will he make more money or will the revenue not change? ROI just answers these questions. You can read it by a simplified or complete formula. A complete ROI calculation can be quite complicated. It depends on the specific costs of the project and the sources of profit. In a simple version, ROI is considered as follows:

    ROI = (Income - Cost) / Investment


    When we consider our entire project from the point of view of an analyst, a game designer, the totality of all the characteristics mentioned here is important to us. That is, the game is considered as a business process, at the input of which there is some main and additional traffic created by the attraction coefficient, there is traffic from the return of the audience, there is a funnel of input conversions that sifts through our players, bringing some of them to the active audience; there is retention, which washes the audience out of the game, and that naturally loses a monthly portion of the players; there is ARPPU and the percentage of paying us money. Cumbersome enough. Both game designers and analysts are trying to operate all this depending on specific tasks. Producers look at the game in the same way.

    And how does the investor look at the game? Is he interested in everything that is told here? In principle, the investor does not care about everything that was considered here. He has his own criterion, "the width of the entrance gate." For a game designer it doesn’t matter how much money you can pour into a project. It is important for him how the project processes this money. Because he does not operate with the amount of funds, he operates with a game. But for the investor, the opposite is true. He has 200 billion dollars. He built Skolkovo, Moscow City or invests in Crimea. He has a lot of money. And your game is able to accumulate and process only 20-30 million dollars. Where to invest the rest of the money? Therefore, only two quantities are important for him: this is the width of the entrance gate, how much he will be able to invest in you, and the ROI - how much he will be able to get back.

    The approaches of the developer and investor are fundamentally different.

    I will be glad to answer all questions about the use of statistics in game design in the comments.

    Also popular now: