Face Meng pm: how to use statistical tools to improve user experience

Recently, the "Meng da" avatar is a hot circle of friends. After 90 s, the Avatar is replaced by a sports baby wearing a German server. Face Meng app was launched at the end of last year, but it was three months after it was launched. In the meantime, what did the operations team do? Umeng interviewed face Meng's product manager to explain the operation story behind the popularity of this app from the data statistics layer.

Face Meng has been using umeng's statistical platform since November 2013 to effectively make product operation and promotion decisions in terms of user needs mining, user experience improvement, and product revision. How do they deploy the various functions of the statistics platform and guide their operation?

I. Make full use of custom events to gain a deep understanding of user Usage Details

 

In addition to viewing daily data (new users and daily active users), the most frequently used event is custom events to understand the usage of users in applications. This function is easy to integrate. You only need to import the SDK package and add a line of code to add a point. It is also convenient to view data in the background.

When the sharing feature was initially added to the product, the team expected users to share the puzzle to public social platforms such as the circle of friends. However, the tracking data of custom events shows that 70% of users prefer point-to-point Sharing. For example, Xiao Li's profile picture can be directly sent to QQ friends or friends. The proportion of social platforms shared with friends and QQ space accounts for only 20%. Through this data, the face Meng Team clarified the product improvement direction and began to consider optimizing the point-to-point sharing feature in product planning.

 

Second, focus on data from various channels and focus on high-quality Channels

 

The Operation Team compares the advantages and disadvantages of each channel based on multiple indicators, and selects the main channel for cooperation. With the new feature "channel comparison" of umeng, you can easily compare the data of Multiple Indicators of the channel on the same page. For example, if the duration of a single application is high, you can find out which channel the user stays in the application for a long time.

Through the proportion of old users, you can learn about the components that different channels bring to users of applications. If the percentage of old users is the highest, but the retention rate is moderate the next day, we can guess that the channel has fewer new users and the quality of new users is average. On the contrary, the low percentage of old users and the high retention rate indicate that these channels provide a large number of users to the app and the high quality. Of course, we also need to consider whether the investment from each channel receives the corresponding return.

 

Third, integrate error analysis to improve R & D Efficiency

 

At the technical level, R & D prefers error analysis. This function is integrated. Once a user encounters a crash error when using the app, the error is reported directly to the app background. R & D personnel can view the problem and fix it in time. According to the experience of face Meng R & D personnel, the error rate ("number of errors/number of start times") published by face Meng is normal at below 0.03%. If the error rate is higher than 0.05% at any day, the packet may be faulty.

In addition, the error details, model, and system functions are also very useful. You can view the error log for troubleshooting. The error model and system provide more convenience for testers. Currently, the anroid device test cannot cover every model. By using the wrong model and system data, you can determine which models and systems are prone to errors and conduct targeted tests, improve R & D efficiency.

The yuemeng product manager talked about the experience of face sprout cloth version April this year in 2.0. At that time, the product was greatly modified, but because there were not many models tested at that time, no device in the anroid 4.0 system was found for testing, and all the devices on hand were released after testing. The new version was released early in the morning. In the morning, technicians found many errors from the error analysis data, most of which were errors of the android 4.0 model. They immediately bought two identical devices for testing, it was found that the anroid 4.0 system had poor support for the new feature "Web webview controls". Many Android devices in version 4.0 could not open the page, and then made urgent repairs and re-released them, to solve this problem.

 

Fourth, pay attention to user feedback and provide reference for product planning.

 

User feedback from umeng is the main basis for product planning. Many users have reported that they want to combine themselves and their friends in the "double" mode. In February June, face cute released a version that includes the "double" mode. Users can combine their pictures. After the "double" mode is released, more materials and the history of previous puzzles are needed. Face Meng is currently preparing to release such a new version and optimize some detailed functions based on user feedback.

Guo Lei, founder of face Meng, said in his report that face Meng's avatar mission has been fulfilled. The team will focus on R & D products and strive to retain users for a longer period of time. Wish face cute team go further!

 

| View umeng statistics demo | SDK download |

 

Note: This article is based on the interview records of umeng. If you are interested in using umeng products and functions and are willing to share it with you, you are welcome to add umengcom.