In operation, it is essential to calculate the churn rate of the number of users, but many people just know the data and cannot find a way to analyze the reasons for churn or put forward strategies to reduce churn. Today we’ll talk about how to do churn analysis using App retention as an example.

01 myth article

First, we need to know two common mistakes in churn analysis:

** First, try to retain every user. ** strong twist melon is not sweet, the loss of many users is unavoidable, such as the transfer of user interest, such as the user does not match the target audience of the product, these users spend more money, do more optimization may have no use, for this part of the user, simply give up, to avoid ineffective investment.

** Second, try to understand each loss. ** For example, one day the churn rate suddenly more than before, these occasional fluctuations belong to the normal situation, in case many users are in a bad mood that day, if we have to do the analysis so detailed but not easy to catch the key factors. For us, it’s just a matter of keeping the attrition rate within an acceptable range.

02 thinking article

The second is the idea of user loss analysis. Our purpose is nothing more than to find out the reasons for user loss and put forward suggestions for improvement. Here, we divide the reasons for user loss into three categories and give corresponding countermeasures:

1) Systematic reasons

If the overall loss rate of the business is lower than that of the peer, it is likely to be a systematic reason, mainly because the business is worse than that of the opponent. For example, the loss of Microvision is higher than that of Douyin. In this case, funnel model can be introduced to analyze the problem layer by layer, and then corresponding optimization suggestions can be put forward.

The retention of users in App can be divided into three stages: entry stage, growth stage and maturity stage, which need to pay different attention to.

In the initial stage, we should focus on user activation, improve without discrimination, and constantly optimize the process of new user experience, so that new users can feel the value of the product quickly and at low cost.

In the growth stage, users need to be treated differently according to the value of users. For non-target audiences and fleece party, if they lose it, they will be dismissed as wasting resources. For target audiences and heavy users, special attention should be paid to activity and attrition, and real attrition often occurs when users are inactive. At this time, the behavior of core users was analyzed, and the details with large losses were carefully investigated and optimized.

In the mature stage, it is necessary to think about the long-term value of the product to users. The purpose of analysis is to insight into the potential related needs of users, actively meet more needs of users, and repeatedly let users experience the value of the product.

2) Event-type causes

If there is an abnormal short-term fluctuation in lost data, it is likely to be caused by an unexpected event, such as outage, price increase, negative news, competitor suppression. User attrition due to event causes usually shows up in the activity rate after the event occurs, and it takes some time for the attrition rate to rise. Therefore, user churn analysis should not only pay attention to churn rate, but also monitor the active rate.

In the analysis of this kind of loss, after the anomaly is observed, relevant negative events should be collected first. Then the hypothesis argumentation, find the core influencing factors; Then find the affected user groups, preferably labeled for observation; Optimize and adjust according to the influencing factors, and observe the change of user activity rate; Observe whether the attrition rate returns to the normal level after a period of time. If not, repeat the previous step optimization.

3) Trend-oriented reasons

Attrition rates that fluctuate within acceptable limits should be ignored, but if they continue to drift slowly higher or lower, they should be of concern. This kind of change sometimes even business people can not explain clearly, we can not analyze anything even with the previous two reasons, at this time we should consider whether it is the cause of the trend.

For example, content e-commerce platforms are slowly robbing the market of X Bao. This trend is slow at the initial stage, and if it is not sensed in time, the consequences may be subversive. However, the churn caused by trends is more difficult to identify and solve than the first two reasons. At this time, if you have exhausted the existing analysis methods and have not found clues, you should consider looking at industry data analysis reports and setting up several observation indicators to track them first, and then you can see the situation clearly if you have accumulated enough.

03 tool post

User loss analysis is done in the end, it is best to use a professional data analysis tools, and then take the mainstream FineBI advantage is efficient and useful, built a large number of analysis model and equations, like the retention rates of these common indicators with built-in formula calculation can be quickly removed from a lot of data processing steps, very convenient.

In addition, FineBI has rich built-in charts. When doing visualization, multiple analysis charts can be put together and the linkage relationship can be set up to create a large analysis screen, which is very easy to use in both analysis and presentation.

In today’s stage of highly homogeneous products, the competition between enterprises is more and more reflected in the competition for users, so it is more and more important to analyze the data of user loss. However, some friends often send me private messages saying that I do not know how to do user churn analysis. I think it is mainly because I do not have a comprehensive understanding of user life cycle. I suggest that I should consult more business or user operation personnel in front of me and make good use of analysis tools.

Analysis tools

Finally, share the analysis tools, click “data analysis” to get the data analysis tools!