5 ideas on game analytics
... which can be gathered from the webinar of our partners from detodev . This webinar was held on August 25th together with friends from the excellent GDCuffs game design portal on the topic “ Analytics in games ”. Sergey Gimelreikh , founder at ORC WORK, Alexander Pashin , Senior Game Designer at Silly Penguin and devtodev leading analyst Vasily Sabirov , who incidentally will soon hold a separate master class on game analytics for students of our program “Gaming Internet Projects ".
Under the cat you will find a summary of key ideas from the webinar, as well as a link to the video.
Alexander Pashin : it is often accepted to evaluate the audience like this: a man, 18-25 years old, with or completing a higher education, income from 30 thousand rubles, and so on. The time of such evaluation is gone, the topic is outdated. Now it is more important to understand that, for example, this user is more inclined towards midcore, towards games of a certain genre. Or this user plays more on some devices, a certain session length is preferable to him. And this issue is solved through the traffic assessment: traffic is estimated, and I can make an impression about the user. And now you can analyze, and I use methods such as funnels and cohorts.
')
Vasily Sabirov : can we unequivocally state that the player should be forced to reach everything himself, reducing the density of training in the tutorial?
Alexander Pashin : I remember, there was an article in which it was shown that a short tutorial with a high conversion in the long run lost in terms of a longer tutorial with the initially worst conversion. It turned out to be less engaging, oddly enough.
Sergey Himmelreich : I ’ll add that it’s a double-edged sword, and doing a long and tedious tutorial explaining every sneeze is also a mistake. I also had a story with a tutorial that I had been explaining for too long, and we, having reduced it, increased the conversion and did not worsen the retention of the project. I also want to add that it is important to take into account technical problems in the tutorial. We analyze not only the user's path, but also the tutorial download, for example, if we are talking about a mobile device.
Real case:
Alexander Pashin : with regard to download speed, it is really very important. The farther from the capital, the worse the Internet. And this greatly affects the quality of the game.
Vasily Sabirov : we had a case. In devtodev it is possible to analyze in detail the first session of the player, build a funnel for more than 100 steps and find problem areas. One of our clients has integrated many steps, including invisible steps, when levels, textures, and so on are loaded. And it turned out that the greatest dumping just happens on these invisible steps - the user does not wait for the download. The developers went to optimize the tutorial technically and subsequently significantly increased retention rates.
Sergey Himmelreich: we must distinguish between classical retention and rolling retention. In the classic - the user is considered retained if he entered the game on the N-th day after installation. In rolling retention, a user is considered retained if he has entered the game on the Nth day after installation or later. Rolling retention therefore gives results higher, and this is often used when talking about retention to investors, but this is not entirely correct. For work tasks, I usually use classic retention. For mobile games, I measure it for a period of up to 30 days, namely: 0th day, 1st, 3rd, 7th, 14th, and also 28th or 30th.
Vasily Sabirov: are you considering long-term retention?
Sergey Himmelreich : no for mobile games; The lifetime is usually low, and up to 30 days is always enough. Although, if we talk about the calculation of lifetime and LTV, then it is useful to consider, say, semi-annual retention.
Vasily Sabirov: my practice shows that long-term retention makes sense, and it correlates more strongly with income. Our research suggests that the longer a player is in a game, the first, the higher the likelihood of making a payment on each subsequent day, and second, the larger each next payment. Thus, long-term retention and the audience increases, and increases its monetization potential.
Real case:
Alexander Pashin: they usually say, for example, about runners, that lifetime is there for 2 weeks. Although I think that for good games it can be much more. In particular, in the Subway Surfers runner, it makes sense to consider annual retention. Or if you take large midkornye GaaS (Game as a Service), then there you can consider retention of 180 or 360 days.
Sergei Himmelreich: that, in my opinion, retention may depend on:
Vasily Sabirov : is it possible to analyze the core loop using a regular funnel? Set the desired sequence of steps in accordance with the goals and steps of the core loop, set the time for each step to pass and see how users step on them.
Alexander Pashin : yes, quite. With the help of this schedule you can see that either the mechanics are not working, or people do not perceive them.
Sergey Himmelreich : when constructing such a report it is useful to take into account the levels and number of attempts spent on their passage.
Alexander Pashin : in general, yes, the core loop is easy to analyze. But we must bear in mind that in f2p-games there is such a moment that money compensates for the speed or efficiency of the passing game.
Sergey Himmelreich : it makes sense to divide the funnels, to watch separately how the paying and free players move.
Alexander Pashin : you need to look so that you can not throw a lot of money and immediately go through the whole game.
Sergei Himmelreich : illustrate an example of a case. I am engaged in card games, and in one game we sold already pumped cards. The player buys a pack with powerful cards, begins to quickly pass all levels, eventually loses interest in the game and falls off. That is, we ourselves lose our paying user.
Vasily Sabirov: what events in the game do you track? What player actions go to your analytics?
Sergey Himmelreich: I will share those events that I track:
Alexander Pashin: yes, I agree. In addition, I still measure ARPPU and LTV, for this you need to track the events of all entries and payments. Well, it is important that a set of steps allows you to build funnels for all key user actions: payment, passing a level, and so on.
Real cases:
Vasily Sabirov: colleagues, can you give examples of cases according to the rule “Everything was bad - they applied analytics - everything became good”?
Sergey Himmelreich: Yes, I can. In one of the games we saw that the players fall off quite early. It was a collectible card game, and we, analyzing the behavior, realized that the players have no long-term goal. They simply lost interest in collecting cards. As a result, we, as they say, “hanged a carrot”: we made an award for a set of cards. And retention rates increased significantly, increased by 20%.
Alexander Pashin: I have rather an antagonist case, as it was bad, until there were analysts. We had a game in which there was an internal free market. We realized that our income is decreasing, but we could not understand why. They have been praying for more than three days, trying to understand the reasons. We did not have the opportunity to see how much money was in the hands of the players, we could not see the movement of goods. As a result, we found out that our players duplicated in-game items, they became larger. We learned about the existence of the black market, and realized that some people earned on our project even more than we did. It was a natural detective story.
On this happy note, we end this article. If you want to listen in detail to those things, a brief summary of which we cited above, then here is the complete recording of the webinar:
Also the blog " Management of online gaming projects " will soon be published an article about game design reports and metrics responsible for the economy, balance and state of the game from our supervisor Konstantin Sakhnov ( Kallist ), which is a continuation of his series of articles on analytics. Subscribe not to miss!
By the way, if you want to listen to our lectures live, then soon there will be one of our public lectures in Moscow, which is free of charge. Details and entry here .
Ask your questions in the comments, we will be happy to answer.
Under the cat you will find a summary of key ideas from the webinar, as well as a link to the video.
Success for a new approach to assessing the target audience
Alexander Pashin : it is often accepted to evaluate the audience like this: a man, 18-25 years old, with or completing a higher education, income from 30 thousand rubles, and so on. The time of such evaluation is gone, the topic is outdated. Now it is more important to understand that, for example, this user is more inclined towards midcore, towards games of a certain genre. Or this user plays more on some devices, a certain session length is preferable to him. And this issue is solved through the traffic assessment: traffic is estimated, and I can make an impression about the user. And now you can analyze, and I use methods such as funnels and cohorts.
')
Funnels and cohorts
Funnel - building the process of achieving a specific goal by the user, allowing you to determine the bottlenecks and growth points of the project. A cohort is a group of users characterized by a common property for a certain period of time. In cohort analysis, users are divided into segments and the behavior of each segment is monitored.
Vasily Sabirov : can we unequivocally state that the player should be forced to reach everything himself, reducing the density of training in the tutorial?
Alexander Pashin : I remember, there was an article in which it was shown that a short tutorial with a high conversion in the long run lost in terms of a longer tutorial with the initially worst conversion. It turned out to be less engaging, oddly enough.
Sergey Himmelreich : I ’ll add that it’s a double-edged sword, and doing a long and tedious tutorial explaining every sneeze is also a mistake. I also had a story with a tutorial that I had been explaining for too long, and we, having reduced it, increased the conversion and did not worsen the retention of the project. I also want to add that it is important to take into account technical problems in the tutorial. We analyze not only the user's path, but also the tutorial download, for example, if we are talking about a mobile device.
Real case:
Alexander Pashin : with regard to download speed, it is really very important. The farther from the capital, the worse the Internet. And this greatly affects the quality of the game.
Vasily Sabirov : we had a case. In devtodev it is possible to analyze in detail the first session of the player, build a funnel for more than 100 steps and find problem areas. One of our clients has integrated many steps, including invisible steps, when levels, textures, and so on are loaded. And it turned out that the greatest dumping just happens on these invisible steps - the user does not wait for the download. The developers went to optimize the tutorial technically and subsequently significantly increased retention rates.
Periods and ways to evaluate retention
Sergey Himmelreich: we must distinguish between classical retention and rolling retention. In the classic - the user is considered retained if he entered the game on the N-th day after installation. In rolling retention, a user is considered retained if he has entered the game on the Nth day after installation or later. Rolling retention therefore gives results higher, and this is often used when talking about retention to investors, but this is not entirely correct. For work tasks, I usually use classic retention. For mobile games, I measure it for a period of up to 30 days, namely: 0th day, 1st, 3rd, 7th, 14th, and also 28th or 30th.
Vasily Sabirov: are you considering long-term retention?
Sergey Himmelreich : no for mobile games; The lifetime is usually low, and up to 30 days is always enough. Although, if we talk about the calculation of lifetime and LTV, then it is useful to consider, say, semi-annual retention.
Vasily Sabirov: my practice shows that long-term retention makes sense, and it correlates more strongly with income. Our research suggests that the longer a player is in a game, the first, the higher the likelihood of making a payment on each subsequent day, and second, the larger each next payment. Thus, long-term retention and the audience increases, and increases its monetization potential.
Real case:
Alexander Pashin: they usually say, for example, about runners, that lifetime is there for 2 weeks. Although I think that for good games it can be much more. In particular, in the Subway Surfers runner, it makes sense to consider annual retention. Or if you take large midkornye GaaS (Game as a Service), then there you can consider retention of 180 or 360 days.
Sergei Himmelreich: that, in my opinion, retention may depend on:
- time zones
- weather and season
- technical problems
- devices
- tutorial quality
Game loop analysis (core loop)
Vasily Sabirov : is it possible to analyze the core loop using a regular funnel? Set the desired sequence of steps in accordance with the goals and steps of the core loop, set the time for each step to pass and see how users step on them.
Alexander Pashin : yes, quite. With the help of this schedule you can see that either the mechanics are not working, or people do not perceive them.
Sergey Himmelreich : when constructing such a report it is useful to take into account the levels and number of attempts spent on their passage.
Alexander Pashin : in general, yes, the core loop is easy to analyze. But we must bear in mind that in f2p-games there is such a moment that money compensates for the speed or efficiency of the passing game.
Sergey Himmelreich : it makes sense to divide the funnels, to watch separately how the paying and free players move.
Alexander Pashin : you need to look so that you can not throw a lot of money and immediately go through the whole game.
Sergei Himmelreich : illustrate an example of a case. I am engaged in card games, and in one game we sold already pumped cards. The player buys a pack with powerful cards, begins to quickly pass all levels, eventually loses interest in the game and falls off. That is, we ourselves lose our paying user.
What events to track?
Vasily Sabirov: what events in the game do you track? What player actions go to your analytics?
Sergey Himmelreich: I will share those events that I track:
- for marketers, it will be very useful to track traffic channels, for this we have our own tools;
- first session, tutorial funnel - we analyze it in as much detail as possible to see the players ’blade in the first steps;
- events of entry into the session and exits from it, with the help of this it will be possible to calculate retention;
- conversion into purchases: moreover, I track both in-app purchases (for real money) and in-game (for virtual currency) so that you can analyze the demand for goods and calculate the consumer basket
- attempts to pass the levels - using this, you can calculate the conversion by difficulty levels in the game
- Finally, I always try to keep track of technical problems in order to see and correct them. Here is my basic set.
Alexander Pashin: yes, I agree. In addition, I still measure ARPPU and LTV, for this you need to track the events of all entries and payments. Well, it is important that a set of steps allows you to build funnels for all key user actions: payment, passing a level, and so on.
Real cases:
Vasily Sabirov: colleagues, can you give examples of cases according to the rule “Everything was bad - they applied analytics - everything became good”?
Sergey Himmelreich: Yes, I can. In one of the games we saw that the players fall off quite early. It was a collectible card game, and we, analyzing the behavior, realized that the players have no long-term goal. They simply lost interest in collecting cards. As a result, we, as they say, “hanged a carrot”: we made an award for a set of cards. And retention rates increased significantly, increased by 20%.
Alexander Pashin: I have rather an antagonist case, as it was bad, until there were analysts. We had a game in which there was an internal free market. We realized that our income is decreasing, but we could not understand why. They have been praying for more than three days, trying to understand the reasons. We did not have the opportunity to see how much money was in the hands of the players, we could not see the movement of goods. As a result, we found out that our players duplicated in-game items, they became larger. We learned about the existence of the black market, and realized that some people earned on our project even more than we did. It was a natural detective story.
On this happy note, we end this article. If you want to listen in detail to those things, a brief summary of which we cited above, then here is the complete recording of the webinar:
Also the blog " Management of online gaming projects " will soon be published an article about game design reports and metrics responsible for the economy, balance and state of the game from our supervisor Konstantin Sakhnov ( Kallist ), which is a continuation of his series of articles on analytics. Subscribe not to miss!
By the way, if you want to listen to our lectures live, then soon there will be one of our public lectures in Moscow, which is free of charge. Details and entry here .
Ask your questions in the comments, we will be happy to answer.
Source: https://habr.com/ru/post/309866/
All Articles