Internal Startup or How and Why We Launch Web Games
Our company has a tradition of creating small internal startups, and not just coding to order. This is both an outlet for developers and a chance to try themselves in new niches. So in December 2023, a side project appeared at Magnus Tech, which then grew into a full-fledged small game studio Cheesewhich has already tested dozens of hypotheses and launched three successful web games, which have been played by more than 200 thousand people.
My name is Nikita Apollonov, I lead the development team. Today I will tell you where we get ideas for our projects, how we learned to work with Russian gaming platforms and why web games are Great Again. And to start, a little about the current state of affairs in domestic game development.
Why Russian platforms for web games are promising and why they are easier than mobile games
The mobile gaming market's decade-long growth, which peaked in 2020-2021, has slowed down. There are several reasons for this:
Increasing Game Marketing Spend and Declining ROI. According to report Sensor Tower, attracting users to the game with advertising has become expensive, while the growth rate of revenue from users (LTV) is now lower than the growth of the cost of acquisition (CPI). Thus, the investment cycle of projects has increased, and the marginality has decreased.
EU regulatory practiceswhich oblige personal data owners (developers) to increase expenses on infrastructure related to data storage and processing. This also includes Apple's default ban on access to IDFA, a personal advertising identifier. Together with the first point, this greatly reduces the efficiency of user data analysis, which leads to a decrease in business margins.
Markets have a strict policy towards developers. The main complaint of developers about Google Play and App Store is the very high commissions for in-game purchases – 30%. Attempts to conduct transactions through alternative payment systems are punishable. A good illustration of dissatisfaction with such a high commission is Epic Games vs Apple lawsuitwhich lasted 3.5 years. The case ended in nothing, as the US Supreme Court rejected the claims of both parties.
Rising costs of game development. Over the past 5-7 years, the work process has become much more expensive, because the cost of work of specialists – programmers, artists, game designers, analysts – has grown. The cost of infrastructure has also increased, administrative costs have increased. The cost of developing a casual mobile game can range from 100 thousand to several million dollars.
Thus, development became more expensive, the investment cycle increased, and the probability of success decreased. According to research companies SuperScale83% of launched mobile games fail within three years, and 43% are stopped in the development stage. After a rise in the Covid year 2020, the global mobile gaming industry demonstrates near-zero market growth rates.
Due to the fact that work on international gaming platforms has become more complicated, some large foreign companies have closed their offices in Russia and, in principle, stopped focusing on the domestic market, and the growth of web platforms has accelerated, we have great opportunities to work for the Russian audience.
Yes, competition is also growing, but the web today is a good alternative for small teams of developers who value freedom and variability for promoting their game projects. And this is exactly the direction we decided to pursue.
Game “2248 Puzzle”
Genre
Our latest release is “2248 Puzzle” Conceptually, this game is a descendant of the already classic “2048”.
In our variation of this puzzle, the player needs to combine tiles with the same numbers or numbers twice as big as the previous one. Swiping the tiles results in one new tile with a bigger number. The goal of the game is to score as many points as possible and create a tile with the highest number. It seems like nothing complicated, but gradually the gameplay creates situations in which you need to use your brain. At least if you want to set a new record.
Puzzles are interesting to us because in this genre:
the audience size though gives in heavyweight genres (shooters and strategies), but still quite large;
Long-term support and development of the project is possible. We are not very interested in one-day games that stay in the top for several days or weeks.
You can try many different mechanics.
Puzzles provide scope for the imagination of game designers and developers, but at the same time, it takes less time to create a project compared to more complex and large-scale games.
Idea and audience
When developing “2248”, we wanted to find a balance between complexity and relaxation for the player. We tried to create, so to speak, a not very obligatory game: a project that will help, on the one hand, to stretch your brain, and on the other – to while away the time, for example, in a queue at the clinic. In addition, the player will be able to beat the record, get a higher place in the leaderboard and a portion of endorphins to boot.
Game concept is not original, and that's okay. People who are far from the gaming industry, as well as novice game developers, often think that in order to break the top, you need to create something unique and previously unseen. This is quite a difficult task, because to one degree or another, everything has already been invented: there are established genres, hundreds of subgenres. Nowadays, what distinguishes a good game from a bad one is not one hundred percent originality, but a successful combination of gameplay and artistic component.
Today you can't say: “I came up with a cool mechanic, and it's going to take off, and I'm going to make billions, because no one has ever made a game like that.” It was either made or not made, because it's complete crap.
Most modern game developers are followers of certain trends. We are too. Everyone is engaged in creative borrowing, but not everyone admits it. We closely monitor existing projects, analyze why some of them are successful, while others are not.
Using such analytical platforms as Sensor Tower, Data.ai, I make weekly selections of mobile games (new and old), look at their metrics, number of installations, how much they earned. I play, look for their analogues among web games and evaluate their popularity there. The popularity of a mobile game does not guarantee that it will be popular on the web – audiences and preferences differ. In general, we research the market and select projects, trying to interpret them in our own way, add something, remove something, and so on.
We also carefully approach the analytics of our projects: we look at how often users open our game, what moves they make. Based on this data, we draw conclusions about what can be changed.
“2048”, which inspired us, has already had over 70 million downloads since its release in 2020 – this is a good indicator. Due to the large audience within the “tile puzzle” genre, many similar projects have appeared: for example, X2 Blocks. It would seem that the train has long since left, but we are not trying to repeat such a Success Story and are not competing with the original project – we have a different niche: a Russian web platform, where foreign companies are not currently represented, as well as a different artistic and technical execution.
Development and publication on the web platform
The development process at Cheesy Cheese resembles that of large game studios. We have people from such companies among us, and they brought with them the usual processes, but adjusted for our modest scale. Development can be divided into 2-4 stages. True, some games do not survive to stages 3 and 4, but more on that later.
Step one: finding an idea for a game
Renowned game studios usually specialize in one or two genres and are well-versed in trends. As a rule, they have no problem choosing what exactly to develop and what project to launch.
We, on the contrary, are looking for a niche, so there are no genre restrictions, so to speak. While working on an idea, we sift through dozens of industry reports and compile reports where we analyze the games we like and the most popular ones. We think from this position: can we do something similar, how difficult will the game be to develop, what competitors does it have, and so on. As a result, a shortlist of ideas is formed, from which the game designer and I select the most promising ones for the development backlog.
Stage two: development
Having chosen an idea, we create a primary technical task — a Game Design Document (GDD), which describes the visual component, game mechanics, object interactions, and each feature of the game. The document is used by both artists and programmers.
The game designer describes the visual component of the project so that the artist understands the idea, and also defines the setting. To do this, we attach references to similar games to the description, provide links to videos with gameplay. Based on the game design document, we create a list of functionality and features, which we give to artists and programmers.
The next step for the artist is to draw a prototype of the game scene, which consists of different elements: for example, in “2248” it is the game board, individual tiles, animation. For the programmer, these will be tasks for implementing the game logic.
The artist's work is divided into three stages: characters, environment and UI. For the development of the graphic part of the game, for sketching (first drafts), our artists use neural networks to a limited extent: they help in prototyping. For example, the game needs different versions of the same object, say, an apple. It is difficult for an artist to draw 10 completely different apples on the fly, from which we will choose one or three pieces, but Stable Diffusion can, the main thing is to make a detailed prompt. This approach greatly increases the speed of content production and decision-making. I think this is the future of game development.
The second stage ends when we release a prototype with the main game mechanics to test the audience's interest in this kind of game. We spend three to four weeks testing it. Given the small size of the team, this period is optimal for us. Then we decide whether to develop the project further. This is how we test many hypotheses and, through trial and error, bring interesting projects to the market.
I would like to point out that the specifics of the gaming industry in general and development in particular are such that it is most effective to use iterative practices. However, beginners often think that if they develop a complex game for a long time and painstakingly, then it will become popular and bring in big profits. Our experience shows that such an approach is a “castle in the air” that is so far from reality that it is not even burdened with a banal hypothesis test.
You can make a game for a year, and it will not find the desired response. There are plenty of such examples. Look at any large studio: they all have failed projects that spent tens of millions of dollars. Failure is normal. It is simply better to fail quickly than to try for a long time and then see that nothing worked out. A good practice is to launch an MVP, collect user feedback and analytics, work on improvements and launch the next iteration. This way we can try more theories and better understand what works well and what doesn’t. We outsource testing.
Stage three: release
There are three main platforms on the Russian market for hosting browser games with advertising for monetization: VKontakte Games, Odnoklassniki and Yandex Games. We bet on the latter, as it is currently actively developing. Yandex Games has a traffic attraction system, they are improving publishing tools, and have strong support for developers. All this is evident from the statistics. In 2023, the platform paid the authors receive about 30 million dollars, and this figure is growing. But there are also disadvantages. For example, moderation in Yandex Games is long and complicated – sometimes it stretches out for weeks and months, and a game can be “turned down” due to any little thing: either the moderator did not like the description, or the name, or something did not load on his phone… Sometimes it was enough to simply repeat the upload the next day without changes: a new moderator – a different approach. We got a lot of bumps before we learned to pass checks the first time.
And it’s not enough for the project to reach publication: the game needs to be noticed by players, and the competition here is quite high.
Yandex Games has a recommendation AI algorithm (popularly called “Recommender”) that looks at how users behave in your game, how long they play and how often they return. The better the metrics, the more organic traffic the platform allocates. The system takes a couple of weeks to initially evaluate the game, only then does it form a dynamic rating, which is constantly recalculated.
So at the start, while there are no statistics yet, you have to rely only on yourself. More precisely, on the competent design of the game page: description, screenshots, icons – you need to approach any element consciously, from a marketing perspective. Of course, you can use paid promotion, but it will not help if the page design is careless. And for the game to remain popular over the long term, it must be… good.
Sometimes developers are unhappy that Yandex Games allegedly attracts traffic to some games and holds it back for other games. This is especially often heard from novice developers who enter the platform with rather raw projects. The harsh truth is that the better the quality of the game, the more traffic Yandex pours into it, although not immediately. An iterative approach to development, attention to feedback, and regular updates allow this quality to be achieved.
Stage Four: Metrics Collection, Development and Support
After launch, we track:
player retention metrics;
average session duration;
Do players get stuck at any stage of the game?
whether the number of technical errors exceeds threshold values.
Typically, such monitoring takes from one to two weeks. If the initial metrics are satisfactory, we consider the project to be good and plan its development. For example, we increase player retention by adding game entities: bonuses, leaderboards, achievement system.
By the way, an interesting fact: more than a third of our players are foreign users. Moreover, these are not only Kazakhstan, Uzbekistan and other CIS countries, but also countries in Europe, North America and even Japan. If we talk about foreign traffic, it is also present on Yandex Games, and, depending on the game, it can stand out greatly, even dominate.
What results did you arrive at?
Following the release of “2248” we concluded that the first version of the game was successful. The average duration of a game session is 40 minutes, first day retention is 15% (we will increase it to 20%).
This direction has begun a second life (the first was during the rise of flash games). Platforms for promoting games are appearing (foreign: CrazyGames, Poki), a new audience comes, the number of developers and the quality of projects grow. This is facilitated by the lower cost of attracting players to web games compared to classic mobile games, as well as the development of technologies that allow even fairly complex projects to be made on the web.
Another conclusion: based on our experience working with web platforms, we realized that we are ready to cooperate with other teams that have game projects, but do not have the opportunity or desire to understand the specifics of publishing platforms. We are ready to act as a mentor, publisher: to post, promote, optimize, monetize the game.
We plan to work more actively on retention through new daily tasks and bonuses (game currency and boosters). We will also improve the overall feel of the game through graphics and animation.
The studio itself plans to increase its audience, including through the release of new projects. We are currently actively working on two games in the casual puzzle genre and will release them soon.
