Emma Povkhan, Head Of Launch Marketing, Aeria Games On How We Doubled Day 1 Retention

Jamie Giggs

In user-acquisition

April 29, 2016

Emma Povkhan works at Aeria Games and gave a talk at App Promotion Summit Berlin on the topic of How We Doubled Day 1 Retention. The talk covered the best ways in which to increase your app retention rates.

You can find the video of the talk below along with the full transcript.


Hi everyone. I was just told that I can go on and ramble and entertain the crowd while people are still coming back from lunch. I have no idea how, so let me just slowly ease you into the subject and talk to you a little bit about this one case which we ran with one of our games and how we improved retention there. So let me start with introducing myself. I have about a decade of years of experience in the entertainment industry for which I am in the online games business and this fascination with entertainment sector started when I was working as a producer on a live TV show, now this is a funny gimmicky picture, but essentially my whole life in that capacity was behind the screen. I was in the control room, I was sitting in the control room behind the scenes during the live show and my main tools for making a great show were first these TV screens in front of me where I could see the graph live of the ratings of the current show which is running. And my second tool was a microphone, the microphone which was leading to the ear of the host in front of the camera and essentially I could ask her to say whatever I felt was necessary to say in that show at that moment and I could see immediately on the screens how that impacted the rating of the show. Having done that for about a year every single day for several hours gives you a pretty good understanding, pretty good grasp of consumer psychology. What makes people engage with your product, what makes them so interested that they want to back your product and keep spending on what you’re offering on the service or the app or whatever. So naturally this led to me working with the gaming industry and one of the products I was working with, Immortalis, was our iOS and Android game, still is, which got quite great results here in Europe with top two grossing in Germany, top three grossing in Austria and in Franceand Belgium, and quite high up Canada and U.S. store rankings as well as the top grossing app. Now, that was a classic success story and we wanted to build on that, we failed miserably and then we went back up with our doubled retention and this is the story I’m going to be telling you today. First a few words about the company though, we are a global “freemium” games publisher with headquarters here in Berlin, we also have a few offices around the globe with in total around 400 employees and around 80 million users to this day. We work with free-to-play or “freemium” games only, which
you’re sophisticated Europeans I don’t need to walk you through what a free-to-play game is. And among all of our portfolio of games, the most determining feature is that our games are originating from Asia. The game was created in one of the Eastern markets, it did not travel well into the West, so we take the game, we localize, culturalize it and publish it here in the main markets. This is what we do, this is what we did with the game Immortalis as well, so we change completely the art style, we changed communication, we changed monetization strategies, we pretty much changed the whole universe and the whole game feel except for keeping the game engine, so our retaining and monetization techniques are hopefully building up on and improving the original titles. Now, why we chose to do so, because we believe that Asian games in general demonstrate much stronger KPIs. They have a much stronger retention, the average revenue per paying user is much higher and average revenue per user can be 5, 10 times higher than the same app, the same game from a Western developer from a Western publisher. Unfortunately, with this, we know that there is almost no penetration between the two big markets and this is where we come in. So this is where Immortalis was completely re-skinned so to say and we changed the cute anime style into the dark fantasy, which worked exceptionally well with the main western markets, Germany, France, U.S., U.K., etc. We changed a lot of different…we tweaked a lot of communication strategies, we tweaked monetization, but I’m not talking about this today with you, I just want to brag a little bit that we did reach very nice results compared to the benchmarks in the industry. Naturally, we thought, why don’t we take this one step forward? Why can’t we do what King does, why can’t we take the same engine and create yet another game on top of it, why don’t we just create another universe, and this is what we did with Battles and Monsters. And we thought, we will redraw the game, we will publish it in the same markets using our normal strategy and everything will work fine. Well it didn’t, at first. When we published, our customer lifetime value was super low, it was so low that it didn’t really make sense to market the game. And this is saying something coming from us because we are part of the Prosepin Group which is, I believe the second largest media holding in Europe, so we have pretty easy access to TV. The game was not performing as we hoped it would. So, as you’re

familiar with the funnel, we started digging. We realized that if we wanted to get to positive ROI we wanted to understand where the drop-off happens and how we can improve it. And we started from the wider part of the funnel, from impressions, and we saw that retention is our main problem. The smallest KPI, the earliest KPI in retention that we were tracking was the three day retention and we saw that compared to Immortalis, unfortunately BAM retention was super low. It followed with the same trends if you compare conversion rate as well, but as said from the beginning of the funnel we needed to focus on retention first and then that would, hopefully, in our minds improve conversion which it did. With the three day data retention, if it’s bad and it’s the lowest amount of data that you have, how do you do, you dig deeper, you go into it and we realized, quite accidentally after a few days of data mining that if we forget the first day of retention, if we forgot what users are doing in their first day, inside the game then actually it’s not that bad. Actually, users in both games behave more or less the same, so the problem is happening with our users leaving the app very, very early on. So we knew where to zoom in and what we saw is luckily, we were tracking a lot of very granular events to the extent that every single step in the tutorial, every single experience that the user had with the game in the first seconds of the game open, we were tracking. That was a lucky thing. The unlucky thing, as you see here, was that we discovered that more than 14% of our users drop off even before opening the game. That was very unsettling, so we were thinking, “Okay, let’s do the following. Let’s list the hypothesis we have.” We ended up with a very long email thread with something like 20 different hypotheses. We ranked them based on easy wins, and the biggest impact that we think they would have if we tried to fix them. We thought that the client size would be the main one, so we just got rid of the happy music at the beginning, and this was the easiest win ever. That result from 43% drop-off to 17% drop-off, that was quick win within the next few days for us. So we were encouraged to keep on thinking, what else can we improve and we were…we started going from screen to screen, understanding why the user

would drop off at this particular screen. What would be so frustrating or so unclear that this would be the last interaction of our user with the game and we saw that…I’m trying to condense this as much as I can, but essentially it was around 400 different bugs, that we’ve discovered within the first 10 seconds of user experience and one example would be you would get a push notification at the app launch suggesting you something content-wise and if you had on your phone push notifications disabled, then that’s it, you would be stuck at that launch screen. So by fixing that bug and several others we eventually were able to push 84% of our users into the game, which was a great success for us. We also thought of any other option that we had to improve the first minutes experience of the user, with removing the loading music, with checking how we can optimize this first interaction of the user with the game, and essentially, for example, another one thing that can be an easy win which can help, when we got the original Japanese game, it was quite messy and it had quite heavy UI. As you probably all know, Western eye is not that used to getting that much information from one screen, so in Immortalis we simplified it. It turned out that’s not enough if you go into more casual art style, you need to simplify further, so we ended up with a much cleaner screen with one single button for user to tap on. Well I say one button, but there is still this skip tutorial which led user right into the first screen of the game and we checked and it turned out that those users who did skip tutorial did not retain as well. So after long debates we decided to get rid of the “skip tutorial” button completely. All those little tweaks helped us increase the first day retention rate more than double and we were super thrilled, so if I want you to take anything from this little case study today is, I would advise you to keep the whole funnel in mind when you’re optimizing your campaigns when you’re optimizing your user flow and start with the wider end of it because that yields best results. When you’re improving retention, then focusing on first seconds really helps a lot and obviously app size does matter and for one little tip, really if you don’t have bespoke [inaudible 00:13:14] testing solution or if you don’t have a big integration, then our easy fix was to just go into user database ID and separate between odds and evens and that helped us a lot at the very early days and yes, no detail is small enough and a more philosophical phrase, that data mining without hypothesis is a journey without a destination. Thank you.