devtodev: “There was no good course on game design in Russian. Now there is”

Asked questions Mikhail Kapranov, account of the Getloyal retargeting system.

About the company and initiatives

Mikhail Kapranov: devtodev has an office in Vilnius and offices in Russia. Tell us why it was decided to open an office in Vilnius?

Vasily Sabirov: Initially we were only in Russia, in Ivanovo. Then it was decided to transfer part of the office to Vilnius.

Then a considerable part of the industry moved there, including us.

And how is it?

Sabirov: Lithuania has a good balance between European and Russian life. It’s comfortable there. Plus, you adapt smoothly there. You don’t end up in a completely alien culture. It’s about the same.

Now, it turns out, you have two offices?

Sabirov: No, three. There is an office in Vilnius, eight people are sitting there, a large development team is sitting in Ivanovo, and we recently opened an office in Perm. I am there now, and we are planning to make the Perm office an analytics center in an analytical company.

Do I understand correctly that most of your employees are analysts?

Sabirov: Most of the employees are developers.

How is the work going on inside the company?

Sabirov: I and any other analysts form the task, describe it and pass it to the developers. They begin to implement the task in Jir.

We also have analysts outsourcing their services. It turns out that we are engaged in consulting. And it doesn’t always have to do with the devtodev system itself. We can simply act as analysts, third-party consultants.

In addition to working on an analytical system, you conduct several courses, invite people from the industry to lead them. The last course was on game design. Tell me, why did you choose such a topic?

Sabirov: There was no good course on game design in Russian. Now there is.

More precisely, there were courses, but they passed, and they cannot be returned. For example, this was the case with the course from Wargaming. There is also a program at the HSE. But there’s another problem with her. It implies only face-to-face participation.

How can an indie developer be in such a situation, who needs to get a certain set of skills right now?

You can read books, you can watch all sorts of videos and you can make a game. You can still do it all at the same time. This is also a solution.

However, there was no course that was always at hand. And we made it.

About working with developers and metrics

Let’s imagine a scenario. There is a client. You analyze its metrics and see a drawdown at some stage. Do you give advice on what needs to be changed so that there is no drawdown?

Sabirov: Yes, I can tell you something from time to time.

We are positioning ourselves not as an analytical system, but as a service.

If a person is our client, it means that there is a certain chat room in which issues can always be resolved.

Can you name a key metric that is important when logging in to use your analytical platform? How much data is needed at all? I will explain the question with an example. For retargeting to work, a database of DAU and MAU is needed so that there is someone to return. Without which metrics will devtodev not work?

Sabirov: In order for everything to work, two events are enough: “player login” and “player payment”. Only on them you can build a huge number of metrics.

For example, if we know the “input”, then we build a DAU; if we have all the inputs and know how these users came, then we build retention. Then, by enabling monetization, we get ARPU, we get LTV, we get cumulative income, and so on.

On quantitative metrics (audience, online, traffic volume, and so on) we look last. For us, quality metrics are in the first place. It is from them that you can understand the main thing.

For example, even if the project has a small DAU, up to a thousand people, then we look at the share of those paying, find out the average income from one active user, and then find out the ARPPU — income from one paying user.

If we talk about some one metric that I would take with me to a desert island, then it will be a metric of cumulative income: how much money the user brings in for his first n days in the project. This metric, when n tends to large values, turns into LTV.

These are the metrics that we use for initial diagnostics. Plus, of course, retention, as always, is a traditional metric of loyalty and user quality.

Were there any cases when you advised to change monetization and the project showed growth after that?

Sabirov: I can give an interesting example from the field of behavioral economics.

In one of the projects, we saw that the user, when he first enters the store, has to choose from several hundred items. Moreover, in the store, some of the lots were sold for hard currency, some for soft currency.

We have a hypothesis that when the user visits the store for the first time, a choice paralysis occurs. He prefers not to choose anything rather than to choose something.

For this reason, we suggested to the developer to make changes: to limit the choice of players who have entered the store for the first time and have not yet paid in the game to six or seven positions.

After the implementation of our offer, the share of paying in the game has increased.

The need to install the SDK or its size may discourage developers from working with you?

Sabirov: I would not say that we feel serious limitations here. Yes, people ask about the size of the SDK, people take the time to decide whether they will integrate it or not.

But we only see those who have come to us. We do not see the teams that decided not to work with us because of the need to install the SDK. We can say that we are in a state of “survivor’s error”. That is, we do not see the whole situation on the market. Perhaps there are such. However, those who come to us remain.

How often do you update the SDK? Do you adapt to the developers in this?

Sabirov: Of course, we update the SDK, but we try not to do it too often, because each new version necessarily leads to the need to make changes on the side of the game developer.

We have updates a couple of times a year. Before each update, we save up fixes, then we make, upload a new version and ask each developer to update the SDK.

About behavioral economics

We have now touched on behavioral economics. At White Nights in Moscow, you just read a report about this.

Sabirov: This is a relatively new industry. For example, the year before last she was awarded the Nobel Prize in Economics.

Can you tell me about her?

Sabirov: This is an area of economics that is partly the opposite of classical economics. Classical economics tells us that all subjects act rationally. Behavioral economics says, “No, we’re all wrong, we’re all human, we make emotional decisions.”

Behavioral economics studies how to take into account a person’s tendency to make mistakes when forming a strategy.

What example can you give?

Sabirov: The most banal is the price tag of $99.

The difference between $99 and $100 is insignificant for the buyer. However, from the point of view of conversion to purchase, the difference will be very significant.

An exciting and yet applied topic?

Sabirov: Yes. Now I’m very interested in her. I like to develop hypotheses within it, offer experiments, tests, figure out what “behavioral” can be added to the game, what mechanisms would help increase conversion.

In particular, the example of choice paralysis is also from here, as is the term “survivor’s mistake”.

Will you tell me about the mistake?

Sabirov: Easily. We often think that all the data we have is all the existing data in general.

When surveys are conducted in games or in any other services, only active users participate in them. The question often appears in the survey: “What is the probability that you will recommend our service to a friend?”

This is a normal way of measuring efficiency, it has its pros, its cons, but it has an obvious survivor’s mistake. The fact is that only a loyal audience takes part in the survey. She is much more likely to answer the question “Yes!”

That is, they should not be carried out?

Sabirov: Why is that? Just need to adjust.

How?

Sabirov: For example, to communicate with those who left, to find out (from the point of view of the same analytics) the reasons for this, to look at the first session of those who left, what was in their first session or maybe what was not in their first session. Then compare their session with the first session of the one who eventually stayed.

With this, you can catch the very ah-moment when a light bulb lights up in the user’s head, when he begins to like the process. If we catch this moment, then it can later be integrated into the mandatory user flow.

The last question is about returning users using push notifications. How do you use them, what do you say to the user, how do you invite them to return to the game?

Sabirov: The user may not be playing, not because he didn’t like the game, but because he forgot about it. He didn’t like her enough to remember her.

In this case, push notifications after the first day without logging in to the second is a very good mechanism that reminds you of “Hey, I’m still on your phone.”

Again, as for behavioral economics and its application in push notifications, I really like it when applications explain to you why they are needed. This is much more efficient than simply requesting permission to enable push notifications.

Our user segmentation system reacts, among other things, to events that the user has completed or has not completed. For example, we can catch those users who entered the store but did not buy anything.

You can make it so that the system throws push notifications to those users who have entered the store, but have not made a purchase. They may be similar to those sent by online stores: “It seems that you have a basket, and you forgot to buy something there?”

For example, a player wanted to buy a sword. But I forgot. The system throws him a reminder about the sword and immediately offers a 10% discount. It also works well for conversion.

The full version of the interview is on the video below.