Team experience FewReview in the promotion of the app through qualitative feedback. Our company provides motivated reviews and other types of incentivized traffic mobile application in App Store and Google Play.
Customers often ask us a question. “If incentivized traffic is effective, why you do not do the application and do not promote them?”. The question is, in General, is quite fair. We wanted to confirm the efficiency of service by example, and at the same time and look at ourselves through the eyes of customers.
Decided to conduct an experiment. To release the app, promote it through our service to see the result. A few years ago, one of our team members have done the app on Windows Phone, so I decided to create a similar app for iOS.
The app is designed to prepare for the exam in the traffic police. The idea is not new, of course, but the main “feature” of the app is the regime “trainer” in which the user gives all the exam questions in a random order. If the user makes a mistake, then it shows the correct answer, then this question is repeated after 10, 20 and 40 tickets accordingly, and this occurs as long as each time the user will not answer this question correctly. This scheme allows you to focus on the most complex issues and facilitates their development.
Monetization of the application, the following. Users get a certain number of questions for free, the rest unlocked for an additional fee. An interesting task was the selection of the optimal number of free issues, but more on that later.
By the way, we released the paid version of the app, but it failed to gain a decent number of installations. The app went into development, and March 4, 2016 release.
The first thing we did was ASO. For the selection of key words used Wordstat Yandex and AppAnnie, unloaded them in Excel, which analyzed the frequency of queries and the amount of competition, and on this basis chose the right keywords. Immediately after release we started to pour motivated traffic through FewReview.
Users had to find the app for certain keywords, download it, use it and write a review. For the first five days was purchased about 150 reviews. One day we saw a strange picture in the panel developers iTunes.
The number of installations measured in the first days dozens, soared to 15 thousand per day, reaching more than 100 thousand the first week. At first I thought that the system analysts failed — other reasons for such growth, we could not find. The answer was simple — the application got in picering.
At first I thought we just got really lucky. But later it turned out that many of our customers, purchasing traffic at application start, also got in, picering. We got to seventh place in the top free apps. As I said above, one of the main problems was the choice of the optimal number of free issues for the user.
On the one hand, the user had to get a app that he wanted to buy it and get the idea of the regime “trainer”. On the other hand, in the presentation of “Yandex” “the importance of the first session in application” from 2016 the following chart: The vertical axis shows the purchase time on the horizontal. It is seen that in the first hour is the largest fraction of purchases.
Decided to conduct A/B testing to understand which option will be most effective, in terms of conversion and profit. We used Amazon AB Testing Solution, unfortunately, was already closed.
Testing for the number of free issues was conducted in two stages, with different prices and number of available questions. In the first test had four groups according to the number of free issues. 10, 50, 100 and 150. Price levels were three.
75 rubles, 149 rubles and 299 roubles. With the growing number of free issues the conversion of the first grew, then began to fall sharply. If you increase the prices, of course, the conversion rate fell, but the difference between 75 and 149 rubles was very low.
The Golden mean came to the intersection of 149 rubles and 100 free questions. Profit from user reached 7 rubles, which is twice the base value. The second stage of testing it was decided to hold with higher prices. 299, 379, 459 rubles.
Grouping by number of free issues also changed. Now tested 150, 200 and 300 questions. With the growth of the conversion rates fell dramatically, and the number of questions described a parabola. The result is the optimum combination was the price of 299 rubles and 200 free questions, providing a profit of 7.2 roubles per user.
Another factor with the potential to increase conversion, — the window with request of payment. Our team found the following timeline. It reflects the results of a study conducted in the dining room of one of the universities in the UK.
The idea is. The consumption of coffee or tea in itself was free, but visitors could voluntarily pay for these drinks. Above the box where students could put money in for 10 weeks hang different images (they are listed on the vertical axis). The horizontal axis shows the share of paid visitors. It is seen that the image of the eye greatly increased the percentage paid.
We thought that such an experience can be applied to the application. The colors we decided to drop — it was decided to test the image of men vs women image on the background. Both options were compared with the background without photos.
This test surprised us — both were less effective than the basic. The image of the girl worsened the conversion rate by 28% and men — 35%. Accordingly, we decided to go back to the standard option.
Send a private cases, in which you managed to improve (or, conversely, worsen) the performance of the project, [email protected] Interesting experiments are sure to get heading Growth Hacks.