Banal errors when updating the application icon

Selecting an icon for an application is an important part of attracting more users. Even if the application is completely free, without advertising, with good reviews, high marks and beautiful design, the application icon has a big influence on the choice of the install / not install person.

A little less than a year ago, the application had a great design. In addition to the design of the application, an icon was created, posters for VKontakte, Facebook, G + and Google Play groups, all of which are designed in the same style. And if with the design of the application and groups everything was clear. Everything was head and shoulders above the previous version. Then with the application icon, the question arose: "And which icon will attract more users?"
')
The application icon is what can be seen in review articles, on the phones of friends. The name and icon are the criteria by which people search for an application. The task was to find out which of the icons people like more and if it can increase the influx of users of the application. The first thing that was done was a survey in the VKontakte group.

The simplest solution at first glance was completely failed. The question was raised as follows:


The response options for the person who just installed the application and entered the group were incomprehensible, because both icons were new to him. In order to conduct a poll correctly, you need to sign icons with numbers and answer options “Icon 1”, “Icon 2”. But before I understood it, several dozen people voted. Calling a survey not biased is also impossible. Regular users of the application clearly said that the old icon is better. New users wrote that minimalism is cool and beautiful.

In 2015, Google created a handy tool for A / B testing. It was set to compare two icons. A few days later, after the start of testing, a comparison result appeared. He showed that the new version of the icon is less popular than the previous icon.


Findings:

1. When conducting surveys you need to correctly formulate the question itself.
2. As it has been repeatedly proven, it is necessary to choose the right way and place of the survey, so that there is no situation: A survey of Internet users showed that 100% of respondents use the Internet.
3. In some cases it is better to believe the statistics.