When it comes to running a mobile gaming business, serial founder Doron Kagan has seen it all. Kagan founded his first game studio back when we still had to press the number 2 button three times to type a “c.”
Today, Kagan is the co-founder and CEO at Game of Whales, an AI and machine learning platform that helps mobile game developers automate improvements to a game’s monetization experience.
Game of Whales was founded out of trying to solve Kagan’s own challenges and pain points around making sense of in-app purchase (IAP) data for each user in an actionable way at his other business Deemedya—a mobile game publisher he founded in 2010. Last month, we announced a strategic partnership between Fyber and Game Whales. With our new Impression-Level Revenue Data product, we’re now able to provide the same level of data granularity for revenue coming from ads for each user.
In this interview with the Fyber Blog, Kagan, along with Fyber president Offer Yehudai, share their observations on the groundbreaking change to optimizing both IAP and ad monetization experiences at the user level. Kagan details how AI and machine learning technology work to automatically drive more revenue for developers, and both Kagan and Yehudai outline why bridging the gap between IAP revenue data and ad revenue data is such a game changer for developers in the mobile game industry. Let’s dive in.
Scott Reyburn, Fyber: If you had to describe Game of Whales to people who are unfamiliar with it, what would you say?
Doron Kagan, Game of Whales: Game of Whales uses AI and machine learning to improve the entire monetization experience, allowing developers to focus on game development. Game of Whales automatically improves any action related to in-app promotion, churn, conversion, ad monetization, and in-app purchase monetization.
Scott: Why has it been difficult for mobile game developers to understand the in-app purchasing behavior of each paying user?
Doron: Google and Apple don’t let developers make sense of their IAP data on a user level in an actionable way. Mobile analytics platforms also ignore per-user metrics that would allow developers to track IAPs for each paying user.
Scott: I’ll ask you a similar question I asked Doron, but with regard to ad monetization. Why has it been difficult for mobile game developers to understand ad monetization behavior for each user?
Offer Yehudai, Fyber: When it comes to ad monetization today, the reporting that developers can get is extremely limited. They’re usually getting how much money they made from showing ads in their game, in this country, from this ad network, and maybe for this ad placement. So, the gap between where a developer earned money from ads—from country all the way to the user level—is very wide. In this wide gap, a developer can guess and estimate in order to attribute the ad revenue, and there are different solutions to use to guess and estimate, but a developer doesn’t accurately know where the revenue came from. If a developer doesn’t know, then they don’t understand. It’s hard for a developer to build a monetization strategy, if they don’t really know what happened. It’s been like this for years.
Scott: Can you describe the difference between changing the monetization experience for each user through manual segmentation versus segmenting in an automatic way through a platform such as Game of Whales?
Doron: Most developers use Game of Whales for the automation and AI. Although we give the option to manually segment their audience, developers prefer, and we always recommend, to build their audiences using our machine learning technology. The main difference is that with machine learning and AI, the Game of Whales platform can group users into much smaller clusters versus an individual segment with hundreds, thousands, or sometimes tens of thousands or hundreds of thousands of users.
Doron: For example, a single segment made up of IAP whales may have two similar high paying users that have both spent over $1,500 on IAPs and didn’t open the app for 7 days. The first user spent $1,500 in the first 3 sessions, while the other user was a non-paying user for 2 months before making the first in-app purchase. Each user is displaying extremely different purchasing behavior. They’re two entirely different types of paying users. If a developer delivers the same monetization experience throughout the customer journey for these two users, they’ll miss out on revenue. By allowing machine learning technology to analyze and match the right offer based on the behavior of each user, a developer gains a lot of revenue.
Offer: What’s very common when a developer launches a new app is this: They will not show any ads for the first 1, 2, or 5 months, for example. Why? They want to limit churn. They want to enjoy the in-app purchase revenue potential and they don’t know what to do, so they will say “no” to showing ads in the beginning. That’s for the first phase. And then after a certain period of time, a developer will say, “OK, I’ll now show rewarded video, and maybe I’ll integrate highly-capped interstitials because it may increase churn. And I still don’t know if I maximized my IAP potential.” Then, after 6 months or so, the developer may open the floodgates, showing mobile ads to everyone.
Offer: Developers manage their monetization strategy like this because they don’t know. We all know that the percentage of paying users in a freemium game is in the single digits. And we don’t know how those paying users behave. The game changer with Fyber and Game of Whales joining forces, is the ability to understand each user early on in a game, in order to identify paying users and immediately suppressing them from seeing ads.
A developer’s potential revenue increases simply by personalizing the monetization experience for each paying user and non-paying user in real time.
Offer: But for users that are less likely to spend in a game, a developer can start distributing ads to these non-paying users to immediately drive revenue. This way, a developer’s potential revenue increases simply by personalizing the monetization experience for each paying user and non-paying user in real time. Should a given user only get an IAP experience? Only an ad experience? Or some kind of hybrid monetization experience? A monetization experience could change for a user at any point over the course of their lifetime. So if a developer starts to see the behavior change for a given user—for example, stops buying IAPs—they can automatically introduce ad monetization, or vice versa.
Scott: What’s the role of automation? How does the Game of Whales platform apply machine learning and AI to make sense of data at the user level and use it to drive revenue and improve the LTV of a user?
Doron: Let’s start with an example. When a user logs in to a mobile game, Game of Whales’s machine learning technology starts to look for patterns. How does a given user react when he or she loses? Does this user go back to the main menu? Does this user continue to play? Is this user incentivized by winning? Is this user incentivized by losing? Is this user playing socially with friends? Is this user playing a stand-alone game? All of these actions contribute something to a formula, and then what the AI is doing is looking for similar patterns from a wealth of data with previous users.
Doron: So this given user is acting very similar to other users in cluster A72, for example. So now that the Game of Whales system understands that this given user behaves like those in cluster A72, it builds a prediction model to predict what this user might do next. Based on the prediction model and current behavior of this user, Game of Whales then runs its machine learning model to learn what is the best offer to offer this user to prevent churn, optimize conversion, or increase average revenue per paying user (ARPPU). After sending the offer to this user, the system will compare the result of the interaction with a control group to understand if it was successful or unsuccessful to this specific cluster. And at the same time, the system will keep tracking the given user’s behavior, because if it matched this user to cluster A72 and now this user changes behavior, the system will now think this user is more similar to those in cluster B22, so it will move this user to B22.
Game of Whales creates advanced groups of user based on their real-time activity called clusters, which is very different than segmentation.
Doron: Game of Whales creates advanced groups of user based on their real-time activity called clusters, which is very different than segmentation—where users are segmented based on what they’ve done until yesterday. In 30 seconds or 1 minute, a user can move among clusters a few times, because the Game of Whales platform thinks that a given user is in one cluster, and then moves that user to a different cluster. Game of Whales is all based on real-time user behavior.
Scott: Take us under the hood of Game of Whales’s AI and machine learning technology. Can you explain to our readers how it works?
Doron: The Game of Whales platform incorporates a branch of AI called the genetic algorithm method. Basically, the algorithm simulates Charles Darwin’s theory of evolution by natural selection, coming up with something that can be equivalent to genes—all kinds of predictions, assumptions, etc. Let’s say a genetic algorithm consists of 100 genes, where, in one day, the algorithm terminates 50 genes, and keeps the 50 superior genes in order to form a new, more effective generation of genes. Now the algorithm has produced a new generation of 100 genes that are better than the previous generation. Every day, the algorithm becomes smarter because the genes that didn’t perform in their own predictions to improve revenue will be terminated in favor of the more effective genes.
Scott: Today, only a few of the top mobile game developers can actually develop their own machine learning framework to automatically change the IAP and ad monetization experiences at the user level. What does a tool like Game of Whales mean for the 99% of other developers out there?
Doron: You’re 100% right. It’s not even a few developers that can really afford the AI and machine learning resources, because it’s not only maintaining the technology, it’s employing a lot of people that need to maintain user segments on a daily basis.
Doron: Many good games from smaller studios aren’t performing in the app stores because the top game companies have the resource advantage to optimize the LTV of each user, taking all of the best users to their games—since their LTV is superior. A smaller studio that doesn’t have the resources to optimize their user LTV will just lose on the user acquisition front, will not have enough users in their game, and their game will die a natural death.
Doron: We’re leveling the playing field by providing all developers with AI and machine learning technology to drive more revenue. Every developer will be able to surface user-level LTV. With these insights, developers can do more UA and compete with the bigger companies.
Offer: What’s interesting about Fyber and Game of Whales joining forces is that Fyber knows ad monetization very well and Game of Whales knows IAP monetization and user segmentation very well. It would take a long time for a developer to try to build their own AI and machine learning tool. Most developers will spend most of their time developing their next title, and optimizing their monetization experience and user acquisition strategy. There are some huge companies like MZ that build a tool like this in-house, but even then, they need to make sure they’re profitable and successfully releasing their next title. Fyber and Game of Whales are heads down, day and night, making sure we’re building the best tools for our developers. And that includes perfecting the best clustering and machine learning tools all of the time. Not every developer is capable of doing the same.
Scott: Why is bridging the gap between IAP revenue data from Game of Whales and ad revenue data from Fyber such a game changer for developers in the mobile game industry?
Doron: In the last couple of years the mobile gaming industry saw an increase in ad monetization, not only in the revenue growth, but also in respect. A few years back, monetizing with ads was considered a bad thing. Developers believed that they should monetize with in-app purchases, and if they were earning money with ads, then they wouldn’t be successful. But now, since the rise of hyper casual companies like Voodoo and Ketchapp, we see ads becoming a very significant part of their business, and even companies that didn’t show ads to their users, like King for many years, started to show ads in their games. So ad monetization has come to the forefront of the mobile gaming industry. And developers that once only needed to manage their in-app purchase revenue, suddenly have a big chunk of revenue coming from ads—however they have very little understanding and control over how they make this money.
Doron: There are many solutions out there that basically just take ad revenue per country, divided by the number of users, and give an estimation of how much a U.S. user generated, for example. With this kind of data, you know that every daily active user in the U.S. is worth 5-10 cents in ad revenue. Fyber’s ability to track ad revenue on a user level, together with user-level IAP revenue from Game of Whales, allows developers to get a holistic picture of their game’s monetization experience. That’s a tool that every developer will need.
Fyber’s ability to track ad revenue on a user level, together with user-level IAP revenue from Game of Whales, allows developers to get a holistic picture of their game’s monetization experience.
Doron: Other companies, from service providers to ad networks, are now trying to offer some kind of hybrid between in-app purchase systems and ad systems. To have one tool that allows you to not only understand the LTV on ads and on in-app purchase for each user, but also to make smart decisions in order to optimize it. For example, defining who’s going to see ads, who’s not going to see ads, at what ad frequency, and to determine it in a smart way based on user behavior. It’s a game changer for the mobile games industry.
Offer: Many game developers eventually want to acquire more users. It’s a game of monetization, followed by user acquisition. When a developer wants to run an effective and smart user acquisition campaign, they need to understand the true value of their users. If a developer only relies on IAP revenue data, they have some understanding of the value, so they can build some sort of UA strategy. When a developer has the full picture of IAP and ad revenue together, and it’s accurate, then a developer can really take their UA strategy to the next level and intelligently acquire users. Just like we have IAP whales, we have ad whales, which are users with long game sessions, where they engage with a lot of ads. Developers want to identify these ad whales. They want to create look-a-like audiences of their ad whales to find and acquire similar users. By running more effective user acquisition alongside intelligent monetization, a developer’s ARPDAU constantly increases.