1 free read left. Create a free account for unlimited reading + AI research.
Create free account[Music] okay so I'm guy and I'm working at a company mobile game publisher and developer called clorium we're very fairly large publisher and developer we were founded almost a decade ago headquarters are in Israel but we have offices across the world Russia Ukraine Europe the US and we have over 1200 people working at clorium there's different Studios across the world now these numbers are not just to boast but you just explain exactly what I'm doing at the company at some point we realized we need some kind of interface with the platform holders so we're a mobile first company our main platform holders that
will work with our Google and Apple and we realized that we need to get better at having that kind of internal knowledge base for the different studios and different teams so that's what I do at the company I try to drive innovation through adoption of new technology and you can think of me as an internal evangelist of a new tech I'm basically a part of the research department I let the R&D guys deal with the development and my background which might be relevant later in this talk is from the EDA industry company called cadence design system not sure any of you have heard of it but it's basically going you know a much different direction closer to the hardware device manufacturers so the first problem I had when I got into clorium is realizing how many Android devices there are right Google Play lists around 15,000 devices and does get a game developer it Dre but
you know brings the question which one of these devices should we target which one should we test on for QA purposes and that was actually my first research Pro project at Laurium and back then I didn't even realize how big that thing is and what I also need realized is that you know you're going to sell a notion of Android devices you need the right tools not sure if that's the best analogy because it's the Mediterranean but that ship isn't going anywhere regardless right on iOS luckily it's like not you know it's not a better focus but don't I always at least this aspect is easier you have smaller device fragmentation and in general 2018 devices will be more powerful than 2017 devices because it's always high-end devices what's even greater is that you get fast adoption of new OS releases so this is from Mixpanel reports back when iOS 11 was released there about the week after
the release you could see that almost all iOS users were using either iOS 11 or iOS then you don't need to support more than two three versions you know backwards tops and when you compare them to you know to the eye to the mobile ecosystem iOS devices tend to be quite powerful so you usually won't have this kind of performance issues if you're targeting the mobile ecosystem as a whole if you want things to complain about you can complain about aspect ratios or the notes or stuff like that but that's not the topic of this conversation life is easier on iOS so I will mostly touch on droid here okay so it brings back the question how do we measure you know the Android ecosystem how do we measure what we should be targeting you can use internal data that's only if you collect that kind of data in the first place and even if you do that skew towards your current clear
base so let's say we're making a strategy game next months we want to make a casual game it's not the same user base so our own data might might be a little bit skewed right instead you can try to go for publicly available data the bad news is that some sources that used to publish this kind of information are dead now open signals used to share reports you nearly analytics had publicly available available Hardware stats you can still work with Unity analytics now but you need to partner with them and implement analytics in your game if you want to get access to the kind of information luckily Google also gives you some Android distribution - with dashboards I'll go over some of these stuff but first I want to show you the problem that you will still have so even if you find the top ten devices in the market right now they still represent only 15% of the market so these numbers are you know going way back but every year you can test it again and you will see that the top devices are just a very small share of
the market so you need to take a different approach unless you want your QA guys to test the game on you know hundreds of different devices and that's basically what I did I'm not sure if you can see all of these devices but they have a few things in common right so you get your Samsung's and your LG's HTC is Motorola's but all of them are running the exact same screen resolution 1440p all of them have a four gigabytes of RAM and all of them are running the exact same Qualcomm Snapdragon chipset with the exact same GPU which is a train of a 530 okay so Samsung is special some of their Galaxy our devices are running on their own Exynos SOC others are running on Qualcomm but except for that all of these devices are basically the same when you look at the engine and not you know on the exterior and all of them are or at least used to run Android 6 marshmallow when they were released and
that's good news for us because what you can do is take the system on a chip approach SOC approach as I call it and then you realize that Android devices are 98% based on ARM architecture the rest are Intel but it's just you know in significant numbers and there's only a handful of hardware manufacturers that actually work with arm to implement that architecture in their chipsets right so you have Qualcomm Samsung MediaTek Huawei and VDI imagination and maybe if you in significant others that's still better than working on 15,000 you know distinct Android devices even if you factor that to devices will have different screen size RAM size Rose OS version that's still better you know it's better for you to look at it from this approach and I'll demonstrate ok so a GPU is basically the thing that will be blocking your graphical performance in
your game and again I went to unity analytics this is you know from way back but what I found back then was that despite having so many different Android devices ten different GPU models represent 80% of the entire android market okay entire ecosystem in terms of units okay it changes over time but the eight ten ten devices to 80 percent ratio still remains almost the same okay this means that if you're picking up ten distinct phones that each of them has a very specific GPU from this list you can cover 80% of the market for QA and that also begs the question wait which device of these ten should i target so what you want to go for is usually the lowest common denominator you want to go for the weakest device because if the game will run on the weakest device it will most definitely run on the most powerful more powerful ones
once again you have benchmark websites that actually test this kind of things every device they let user test their devices they give them an arbitrary raw score but what should matter to you is that if one device got four thousand point and the other guide eight thousand point you target the device which got less points and then the most part and the more powerful device will be able to run your game as well that also applies to OS market segmentation again I mentioned Mixpanel earlier they show you iOS fragmentation but for Android fragmentation you could look at official numbers from androids own distribution the dashboard ok you might want to run a sanity check these represent the entire ecosystem and maybe not the you know game segment so you got Android devices running you know TV set tops and tablets and phones so you need to ask yourself is this actually representing my market but it's still a
good rough estimation but it's you know it's one point in time they will show you the devices as of right now and how you know how the OS versions the distribution is among these devices but you're not targeting the game right now you want to release your game next year or the year after that and that's when you can do some data manipulation how can you predict the future to quote a good friend of mine person known as Kane he who controls the past commands the future and what you can do is use web archives anyone familiar with web archives okay so Google's own dashboards have snatched a snapshot from earlier years so you don't need to be a statistic it is the statistical genius to plot some of these numbers from earlier and then you can see where each OS version is going so even if you target you know want to use into the future you get a kind of ballpark of where the OS is
going so you can say that hey two years from now I won't have to support Android 5 anymore and which six will most definitely be the low you know the lower lowest OS version that I need to support and lastly but perhaps most importantly is the Google Play console device catalog how many people are using that right now okay only a few more of you should you know if you have a game on a Google Play more if you should look into that because it's a very powerful a resource for realizing which devices are running your game so again it's a little bit skewed to your towards your own games already available in the marketplace but it still get great resource because it lets you exclude games from devices that you don't want to support so for example let's say we found a very specific bug in the Nexus 9 device we can cross this game out so
Google Play users will not download it and realize that the game is broken but if we go back to this OSI approach which I mentioned this tool also lets you filter devices based on this kind of specs ROM screen size SOC so instead of just removing Nexus 9 let's remove all the devices that are running a Tegra k1 by Nvidia which is the SOC powering the next nine and a few other devices so again that still a bit technical but what's even better with the device catalog is what you can get from a marketing perspective right so you have a technical question your developers are coming to you they're saying that it's really hard to support the game some devices would lend less than one gigabyte of RAM should I do it and the marketing question is actually what percent of my player base is using devices with less than one gigabyte of RAM and even more interesting question is how is my game rated amongst this kind of devices maybe I have many
players but the game doesn't run well on these devices so I'm getting really bad reviews and the nicest thing is what revenues so a percentage is coming from this kind of devices now Android gives you at the Google Play dashboard for Android they give you all this information these numbers are fabricated by the way we're not making three billion dollars on Android alone in 30 days at least yet and but basically that's what you can get for each segment each way you segment your Android devices on the Google Play dashboard right you can get the install and the ratio out of the total you can get the cumulative average rating compared to the total and the revenue spent compared to the total so now your marketing guys can also look at this kind of technical information and give a marketing answer for the developers and that's extremely important if you want to make the right business decisions again it's awesome if
you haven't been using it already start using it but then again what I also mentioned earlier you have to remember the market changes over time this data is slightly skewed what's correct for your game right now might not be relevant for the next game you'll release one or two years into the future but with that in mind again very useful and something you should always have as part of your toolbox okay um that's it I guess [Music] you
Mobile game developers face significant challenges regarding device fragmentation, particularly within the Android ecosystem, which features approximately 15,000 distinct devices. While iOS development benefits from a more unified hardware landscape, rapid OS adoption, and higher-performing devices, Android requires a strategic approach to quality assurance and hardware targeting. The primary objective for studios is to identify a representative subset of devices that ensures broad market coverage without the impractical necessity of testing on every available handset.
To manage this complexity, developers should shift from device-specific testing to a System-on-a-Chip (SOC) and GPU-centric methodology. Because the vast majority of Android devices rely on ARM architecture and a limited number of GPU models, testing against a small, curated list of hardware configurations can cover up to 80% of the market. When selecting devices for QA, the most effective strategy is to target the lowest common denominator—the weakest hardware—to ensure that the game performs adequately across the entire spectrum of user devices.
Data-driven decision-making is essential for both technical and marketing teams. Developers should leverage resources like the Google Play Console device catalog to analyze performance metrics, user ratings, and revenue generation across specific hardware segments. By cross-referencing this internal data with historical trends from web archives and official distribution dashboards, studios can forecast future OS requirements and hardware shifts. This integrated approach allows teams to make informed business decisions, such as excluding problematic hardware configurations or determining whether to support low-RAM devices based on their actual contribution to the player base and revenue.
Each row is a game or company this document names, each column a document, oldest first. A mark means that document names it. Select a row to see those documents.
Deconstructor of Fun · 2026
Comments