Founder's advice. Yegor Guryev - on the experience of creating a service for streaming computer games

Almost two years ago, the first cloud-based service for streaming computer games, PlayKey, was launched in Russia. It allows you to play games over the Internet without downloading or installing them. In addition, not all users have the opportunity to constantly update their PCs in order to “pull” the system requirements of modern games.

The founder of PlayKey, Yegor Guryev, told Habru about how the service was created, what problems stood and stand in the way of its development. He also gave advice to beginners and to all those who would like to create an IT project on their own.
')
Egor, are you the only company founder?

I am a co-founder of the company. There are two more: Alexey Lykov is the technical director. Vadim Andreev is engaged in interaction with copyright holders.

Three of us previously worked at Enaza, engaged in game projects, digital distribution, developed websites, including Zabava.ru

How long does a playkey last? What are your results today? What direction are you developing in?

We started this project in May 2012, that is, quite a long time ago. Our first commercial product appeared at the end of 2014 - December 23. So, until the end of 2012, we were engaged in the development of the platform and spent a lot of time. Technology, in itself, is high-tech, very complex, there are a lot of pitfalls. This is due to the fact that it is necessary to ensure the operation of different games on our platform. A product for streaming one game can be done in six months.

How is your streaming implemented?

The game runs on remote Playkey servers and is processed there. The user does not need to either download the game or install it on his PC. Just play at your pleasure, even on an outdated computer or laptop by today's standards.

Do you give it all as?

We capture an image from the game and stream it with a video stream to the end user. It's like an online video that can be easily managed in real time.

What protocol?

We have our own protocol for transmission. And for coding, we use the modification H264. Therefore, the user needs to download our Playkey client.

Do you deploy a separate server for each new game?

We have several physical servers. We use two options: on one server we simply start Windows without virtual machines, and within the framework of this OS we run ten games on one system; The second option is virtualization: we divide the server into VMware virtual machines, on each of which we also run several games.



You exist for two years. How many users do you have now? What is your main metric at the moment?

Now we have 27 thousand people who paid us at least once: one of them became a subscriber, and someone just paid an hour of play. For us, this is an important metric.

What are the main lessons you can learn from the history of the company, its development to the current point in time?

The most important point on which I would like to draw attention is not to rush to make a product, but to set aside two or three months to better understand the problem that the project should solve.

I have seen a lot of startups. All of them are inspired by their idea, they believe that it will change the world. As soon as it comes to mind, they immediately begin to implement it. Inspired by the desire to change the world, they are not working out enough solutions to the problems of users who can solve their project.

We also had such a mistake. Because of this, in mid-2015, we had to change the direction of development and spent a lot of time rethinking the product.

Did you have a full pivot? Or did it concern some things around the product?

It was a conceptual rethinking. In terms of technology, there may not have been much change. In terms of product positioning, the changes were colossal.



What exactly has changed at this point? Now I see the opportunity to play a game that will not work for me because of system and hardware requirements. What was it until 2015?

Before that, we developed our own gaming service, which was supposed to aggregate all kinds of games — good and bad, old and new — and would allow the user to play on any devices. We wanted to make our own steam, which would work on all platforms. Changes have placed a priority in the direction of gaming innovations, the technical characteristics of which are beyond the power of most ordinary home PCs.

The conceptual rethinking and subsequent modifications affected the cost of the service, the business model, the economy of our project (it did her good), marketing — and practically everything. But the technology in this case was secondary.

Therefore, do not rush to not make such mistakes.

What other lessons have you learned?

Do not make the perfect product and focus only on the most valuable features. You should not dream and invent functionality that looks interesting, but does not solve the main problems in reality. If customers do not have an urgent need for these features, they do not need to be done. The costs for each of them are about the same, and the value of a particular functionality for the client is very different.

Can you single out the most difficult task that you had to solve?

Now the most basic challenge is to select those tasks that are important for the client with limited resources. If we make ten features conditionally a year, and the team has a hundred ideas, we need to somehow choose the ten most valuable to the user.

How to do it? What do you use to understand which features are primary, which are secondary? Are you asking users or are you trying to navigate yourself?

We all ask the users. Personally, I get about fifty messages a week from them. We have a group in VKontakte, which we actively conduct, there are already more than 25,000 subscribers. We do not use it to attract customers, to promptly inform, for feedback - to find out what users want. We do 50% of features only after user approval. We conduct polls: for example, about which games to launch in the first place ...

Of course, there are other tasks related to security, scaling, automation, refactoring, and so on. We can’t ask users about this. Here, by virtue of deeper professional competencies, the responsibility is entirely on us. That is, it turns out a certain "mix".

What advice would you give to a beginner (maybe even a child) a person who plans to move along the same path as you?

Teaching people not to make mistakes is almost an impossible task. There are a number of things that a person understands only when he has already made a mistake. Therefore, the most correct advice, probably, is to make more mistakes. And quickly, because the question of time is critical. Then the person will understand and learn. Life is short, there are few hours in the day, so you should try to make these mistakes faster. Do not try to avoid them. The main thing is to draw the right conclusions from them and move on.