Game analytics: AI is capable of more than you think
Online games is a special field of artificial intelligence application as players leave a lot of data which can be used as the basis for business decisions. Today we’re going to discuss how artificial intelligence may be beneficial for the gaming industry and what tips should be taken into consideration while putting AI into practice.
Every game developer aims to improve conversion and monetization as well as to reach a new audience. In order to obtain that you have to advertise and promote your product, but it is also important to accumulate and analyze the data about players’ behavior so that the game could be developed in compliance with users’ wishes.
Game analytics are familiar with such metrics as emotional state, response to the game mechanics and users’ behavioral patterns. But only personalized detailed analysis of all these metrics for each player allows you not only to increase the number of conversions for each user but also to ignite “word of mouth”, making the game more attractive for users.
Usually predictive analytic algorithms are used for that. But when it comes to this technology many people recall the “Black Swan” phenomenon and claim that these methods are based on the data from the past. This is absolutely true, as predictive analytics allows you to compare the data from the users that you already have and make forecasts for the future. Analytical tools, eventually, are not 100% accurate in their predictions but they do provide the forecast assessing their accuracy level as well.
More than just predictive analysis
Nevertheless, online gaming offers more possibilities to deeper analysis by means of integration of the neural network with the game engine. In fact, gamers leave huge amounts of data that can be analyzed on the fly. The time spent thinking before the purchase, number of entries and exits in each zone, behavioural patterns in different situations, frequency of the app usage, response or absence of response to mailings and push-notifications — all that is precious data for complex analysis. AI is capable of analyzing each user while a person can only make general conclusions about the areas for optimization.
Gamers’ behavior and emotional state
We are currently working on development of machine learning algorithms tracking and analyzing f2p and match3 players’ behavior. That allows the neural network to constantly analyze the reasons of the behavior and track gamers’ emotional state. The algorithm analyzes many factors and helps create personalized offers based on users’ psychological types.
Profiles segmentation
AI technologies help you sort all your users into several segments. It’s up to you to choose how you want it to be done. For instance, you can define 30% of the players who are most likely to purchase anything in the near future, or, conversely, find 10% of those who are willing to quit the game. In that case you can act accordingly: create special offers for each of the groups.
Another option for segmentation is based on social groups or psychological types. It takes into account the gamer’s typical behavior, the kind of situations that are pleasant for the player as well as the type of possible motivation for purchasing. This approach helps you improve ARPU by means of customization of the offers. Thus, Amazon, a worldwide known retailer, is sure that 30% of their sales come from precise personal recommendations.
Mailing bases synchronization
Most developers remind their users of their game via regular mailings or push notifications. However, using the same mailing template for everyone is not really efficient. Developers tend more and more to create different mailings for various customer categories, e.g. for active players, for those who haven’t used the app in a while, for players with low balance etc. The more mailing patterns you have the better the personalization is, but it gets harder to choose the right target group. At Smartcoders we solve this problem by means of customized ads based on gamers’ psychological types and integration with the game engine. Every following action (or absence of action) of the user improves the accuracy of the classification and may lead to transferring the client to another mailing category. Thus, a user who has just purchased “the crystals” will never be offered to buy ones right away — the customized mailing will now have new game goals.
Analysing reasons why players leave
While new gamers start using the platform, the old ones leave it. However, it is possible to slow down the quitting process. In order to do that AI analyzes the behavior of your whole audience and reveals those players who are most likely to leave the game. Based on the data collected about each user, you can find out the reason why they’re about to leave: someone has been struggling with defeating the dragon for a while, someone else never has enough coins, for others the game is too easy and has become boring. In every situation there is a chance to retain the gamers by offering them something they really need right now.
Another situation is a “dormant” user. In order to re-engathe those ones it is necessary to analyze the whole history of their actions and find out what their problem with the game was. Today artificial intelligence is capable of handling this without any intervention of an analyst at all, offering game process improvement as well as personalized mailing for the user.
Game mechanics tuning
Anyway, game mechanics is to be constantly monitored. As we know, getting to the next level is supposed to be interesting and challenging, but not too much so that the player wouldn’t get stuck in the game process. Artificial intelligence helps to reveal “the bottlenecks”, which reduce users’ motivation (possibly, in compliance with their psychological profile), as well as the opposite areas of the game, which are too easy to pass and do not require any special resources from the gamer (i.e. no purchases are made).
For instance, after proving the “chest” mechanics ineffective Indigo Games changed their gameplay and managed to increase ARPU by 20%, and player retention — by 2–3%. Taking into consideration that AI reveals that kind of trouble in real time, based on gamers behavior analysis, the benefit from this kind of analytics may be very impressive.
Development together with neural networks
Although game analytics has been a reality for a while, competent usage of AI allows you to personalize game development more, to improve audience engagement, to retain the players and to constantly increase ARPU. However, all that becomes possible only in case of neural network integration with the game engine.
At Smartcoders, our aim is to reach transparent and partnership based interaction with gaming platforms, that is why we’ve created an SDK which easily integrates with any system via opened API. Moreover, if you have second thoughts about the result or you don’t have the budget to develop analytical tools at your disposal, we can launch a joint project implying payment by the outcome. These kinds of initiatives make predictive analytics affordable even for indie developers.
Try our demo version at our website, follow the link smartcoders.ai!
