Interview with the co-founder of the service to increase the conversion of mobile applications SplitMetrics

In this interview, we talk with the co-founder of SplitMetrics service Eugene Nevgen. The company has developed a service that allows you to test such an important indicator as the conversion of an application page in Apple and Google stores. In addition, SplitMetrics allows you to identify a number of factors affecting the user's decision to install. This allows you to increase conversion by changing the descriptions, pictures or videos about the game. The idea is very interesting, and thanks to it, the startup managed to get an investment, and now the company is opening an office in the USA.

I hope you remember that this is not the first time we are covering this topic. Basic tips on optimal presence in the app stores, we have already published in the blog Appodeal . By the way, do not forget to subscribe.
As for the interview, it is presented in the following format: “Q:” - questions of the main interviewer, “E:” - answers of Evgeny Nevgen. It's simple.

Q: Considering that a substantial part of people already know about you, let's try to tell you very briefly about the company. So, your main task is to optimize landing pages. The first thing that comes to mind to the person who hears about this is the question: “But you do not have access to the statistics of the application store.” What are you doing?
')
E: This is quite a frequent question. So, we create web pages that have the same elements as the App Store or Google Play. This is a web page hosted on our domain, which has the same elements and works on the same principles as a similar page in the App Store or Google Play. It is on these pages that we conduct experiments, and visitors get on them from advertising.

In addition to the fact that we are engaged in optimizing application pages in the App Store, we also provide analytics of visitor behavior, that is, we not only answer the question of which icon or screenshot to put, but also provide information about how visitors of the target url behave . That is, whether users are interested in your screenshots, whether they scroll them, how long they look at them, whether they open full-size screenshots of the gallery or look only at miniatures. Users read the description or not. Depending on this, you can already make decisions and draw conclusions about whether to invest your time and money, for example, in the translation of the description of the application. Or it is better to focus on the screenshots.

The question about the video, video on the page - the so-called app preview is very relevant now. Do I need to do this app preview or not. We have a rather frequent case, when clients come in, ask such a question, create some very simple video, let's say they record a screencast using QuickTime. Then, marketers create an experiment with video and without video and see if visitors are interested, if they are watching this video and how it affects the conversion.

Q: How much traffic is needed to make the right conclusions?

E: It depends on the conversion in the App Store and, in general, on how actively visitors are moving through advertising and how interested they are. That is very dependent on targeting.

Q: Yes, I see. If this is a highly targeted application, it will need more traffic to see some actions.

E: Or vice versa. We often had such cases when the conversion to the App Store was very low because the application was too wide. For example, a case with a specific, extremely clear application, for example, with offline maps, their conversion is very good. In games, the conversion is so-so, because the competition in this segment is very high. In photo applications, since this is also such a broad topic and there are many applications, also a conversion so-so. In general, we recommend to send a thousand to one and a half visitors to each variation. The tests of two variations, respectively, about three thousand. If you have a high conversion, you will need a lot less.

By the way, it is not necessary to buy traffic, let's advertise. Most of our customers buy traffic on Facebook. There the traffic is, of course, very high quality, targeted, but, to put it mildly, not the cheapest. At the same time, there are other solutions, other platforms for attracting traffic, and a sufficient number of our clients use cross-promotion, that is, advertising from other applications for such experiments. It also works quite well and allows you to save on traffic and on advertising.

Q: Are there any limitations, pitfalls? What applications are not suitable for testing?

E: In principle, what I have already said: if the application does not attract users too much, a problem may arise - it will take a lot of traffic to analyze the conversion. Some of our clients initially cannot assume this, launch the first experiment and understand that they need a lot of traffic, but they cannot find that amount. This is probably the only more or less noticeable problem.

Q: What is the usual starting budget for testing?

E: It is difficult to answer this question. I think, from 100-200 dollars and above, but this is very approximate.

Q: Ok, the order is clear. Let's go back to the issues that may disturb the developers in the mechanics of your work. What happens to users who come to the test page of the application? They come and believe that they are in the catalog.

E: Naturally, this is, in general, the fundamental, main principle of the service - the user does not need to know that he is participating in the experiment. When a visitor arrives, he is already allocated some kind of one variation, that is, he cannot see other variations. He does not know what is involved in the test, and behaves as he would have behaved under normal conditions. That is, for him, this is the usual conditions. He studies the page and decides whether to install the application or not to install it.

Actually, this is the experiment: to see whether it installs the application or not. As for the fact that this is not the App Store, this is a web page that is designed so that users do not distinguish it from the app store. We conducted research and noticed that users do not care. They see the familiar elements for themselves, they see the icon, the screenshots, which will scroll in the same way.
There is one thing here: initially, we made other application store designs. That is, the test page was less similar to the App Store and Google Play, but users still behaved as they used to.

The second interesting point: if you buy traffic, say, on Facebook, then on Facebook, the url opens in a web view, in a native application, that is, there is no redirection to Safari, to another browser, and for Facebook users, principle, looks very similar to the directory. If I didn't really understand this topic, I would simply not notice that this is not the App Store.

I also want to say that we do not collect any personal data of users. All data is impersonal and depersonalized. And we do not collect any passwords or credit cards, we do not even have to install the application. We only make a test and then send users to the current App Store or Google Play, where they can make a purchase or install the application for free.

Q: I was just hinting at the last moment. Are there any negatives associated with the fact that the user comes, roughly speaking, in the App Store, and then gets into it again?

E: 30% of visitors at this stage break away. When they did see the App Store again, some leave. Incidentally, I read the study that even when the App Store asks for a password — that the user wanted to download the application, clicks download, they say, “Enter Apple ID or Apple ID password” —that is, about 30 percent of users fall off. Just because an additional dialogue is being shown where you need to do some kind of action.

In our situation, about the same story: when the user just sees again that he needs to do something, some of the visitors leave. Plus, a certain part falls off due to a weak Internet when redirecting to the App Store. But our main goal is not to force the user to download the application in the application catalog, but to understand how to make the user download the application in the real store. That is - to conduct an experiment.

Q: Ok, another question. Suppose I am a new developer and I want to optimize my page. Why should I spend money on A / B testing of my application instead of going and listening to general tips - they say the video is useful, it works for everyone, add it to your page.

Here Rovio, for example, passed with you a huge test. Imagine that they took and released on Habré an article on how to better optimize the representation of their application on the App Store, in their experience, with their statistics. How different will the tips be to different applications from different themes with different target audiences?

E: Such tips, guides, articles are very good for adding to the list of hypotheses, but not for making decisions. Why? Because even Rovio themselves made a great discovery for themselves. For example, they (and most other game developers) have always used horizontal screenshots.



Previously, it was thought that this should be done because, most likely, gamers are more accustomed, it is more convenient for them to study game play on the full-screen. And for other tests of Rovio, we saw that they often used horizontal screenshots. In the second part of Angry Birds, they decided to just check: what will happen if the screenshots are vertical?



And it turned out that vertical screenshots worked better for this game. We also had a certain hypothesis, which is precisely because Angry Birds is a game that even people who are very far from game player can play. For example, I almost always keep my smartphone upright, horizontally very rarely, if only I watch the video.



I want to say that the advice of other people is great, but you need to check everything on your application and on your target audience.

Q: This case is very intuitive, it becomes clear why testing was conducted. Are there any other experiments that led to positive, which can be said, or is it a mystery?

E: Now nothing really comes to mind. We had a test with a major seller of airline tickets. The bottom line was that this company made a video that almost no one watched.

That is, again, the question arose: why? We concluded that this happened because the test was done on Russia, and in Russia many people already know them, and if someone comes to download the application, the user will not be distracted - he needs an application. It is not too important for a user in Russia to watch this video - everyone knows that this is an application for buying tickets. What could be the surprise inside there?

But at the same time it is interesting, the video very often increases the number of scrolls and screenshots. If we are talking about the App Store, the video is put in place of the first screenshot. And visitors squander this first screen to see others, get involved in the scrolling process, scroll further. This user behavior has been confirmed in our tests. Perhaps there are other examples, but we did it that way.

Q: Here I will try to suggest that, maybe, some case will be recalled. In my experience, when testing was conducted, for example, with some content in social networks, sometimes it turned out that some bad content, obviously bad, comes in unusually well. For example, if you put a photo of homemade salad Olivier and a beautiful salad Olivier made by the chef, the first one works better, because the appearance of the “homemade” salad is very recognizable. It does not need to be perfectly beautiful.

Or another example: sometimes the texts written by the caps work better in Odnoklassniki, although they look awful. But such a text takes and works better for conversion. Here, maybe, there are similar examples when some thing that should not have worked, took and worked?

E: It’s not even a separate story that comes to mind, it’s in principle such a frequent approach, when the designer decides to change the current screenshots of the application to more recent, professional ones. He makes minimalistic screenshots: for example, just an iPhone and an inscription on top. The designer draws screenshots on his 27 inches, and makes the font, sexy iPhone, screenshot and that's it. But it often happens that new screenshots work worse precisely because the designer believes that everything should be very simple, very minimalistic. But sometimes this thing doesn't work. Again, this depends very much on the audience. You are talking about “Classmates”. The designer sometimes does not think that more than half of the visitors may not open the full-screen screenshot gallery and they simply will not be able to read the text - too small, it is not visible on the thumbnails, it merges tightly.

But this is not universal advice, this is not always the case. If you look at Instagram screenshots - they generally have nowhere easier. A slice of the interface and text on top. But Instagram is something, it is also recognizable, well-known application.

In general, you need to take screenshots so that when the visitor looked for just a second, half a second, he could catch his eye for something.

Q: If the owner of an application who has not yet done testing reads us - perhaps his application has not yet been released, he thinks about how to launch it. What do you recommend, go straight to you and start beta testing or read some literature first? Maybe you would recommend that you read on the topic of proper design in addition to the publications that can be found in our blog?

E: First, I want to immediately note that we allow testing to take place even before the application is released in the App Store. Here are the main tasks of our service: the first is to determine which icon, screenshot is better, the second is analysis of user behavior, and the third, very interesting part is that our service allows application owners, start-up owners to better understand their positioning.
We had such a case, and not one recently, when customers came to us, they had a product, had an idea for a product, but they did not know how to position it correctly in the market. And for such cases, you can run tests before the application is released, before it is output to the directory. Then when you click on the "download" button in the App Store or Google Play on our landing page, a dialog will appear that the application is not yet available, but you can leave your e-mail and we will let you know as soon as it is released.

On the one hand, a test user base is being assembled before launch, on the other hand, it is already possible to conduct tests. What is very important is to try to understand your audience: what is the audience, what is the targeting here. Before this, it is very difficult to do some tests, because the traffic will be very expensive due to the lack of targeting. Such tests may allow you to better understand which audience behaves and which audience is more interested in your product.

Q: But in order to start the tests, you still need to make a series of pages, that is, you need hypotheses, you need basic options. Would you recommend something here? Where do people go to prepare?

E: You need to show all the features of the product. And the most important thing in the presentation of your product in the application catalogs is to show what benefits this application will bring to the client. Roughly speaking, why the user's life will become better and happier after he has this application. I personally have such an approach.

Q: Well, with this clear, let's look at the potential future. First, congratulations on getting an investment.

E: Thank you!

Q: What are your plans for further development?

E: Oh, plenty of plans. In short, we are building a platform for mobile marketers, not developers. So that they can receive the maximum number of installations in our difficult time, when CPI tends to infinity. We can show how to squeeze the most out of current advertising campaigns, how to squeeze the most out of other channels. That is, it is all sorts of optimization of the process of obtaining users. While the implementation of this project is a secret.

Q: You grow quickly and help customers optimize the representation of each application. Can you cope now?

E: We position ourselves as a self-service platform, that is, we are by no means an agency and, naturally, we have a support that supports our clients. Nevertheless, we do everything so that the user can cope without any problems himself, automating everything as much as possible. The user came at any time, even at night, he himself started the experiment, made the conclusions himself, made the decision himself and uploaded his application in the App Store. Of course, it’s already hard for us, because there are already a lot of customers. On the other hand, we are investing in a product to make it even more understandable.

Q: Great, it was very interesting to discuss these issues. I wish to move forward and hope that Appodeal customers will be able to get the most out of your service.