Three misunderstandings about user churn analysis. Have you made any mistakes?

User loss on a platform is inevitable, and the constant replacement of old and new users is a normal phenomenon of product upgrades.

However, companies can seek breakthroughs from the proportion of lost users and changing trends, create products that retain new users, thereby improving their ability to retain users and identifying product development trends and space.

When we find that the user churn rate increases, we need to conduct a more detailed analysis of the lost users;

However, when analyzing user churn, we sometimes fall into certain misunderstandings without realizing it, and even reduce the effect of user return due to misjudgment.

1. The first misunderstanding: unclear definition

Some people see that other platforms define the user churn period as not logging in for half a year, so they define their own user churn rate as not logging in for half a year; they follow the crowd but never think about why other platforms define the user churn rate as half a year, what are the similarities and differences between their own platforms and others' platforms, and whether this definition is appropriate.

So before we start analyzing lost users, we need to first clarify one thing: what conditions can define a user as a lost user? Is it a user who hasn’t visited the platform in several months? Or users who frequently visit our platform but haven’t made any purchases in a few months?

For example, during the heyday of QQ, almost everyone would register a QQ account, log in, upgrade, play games, chat, and perform other operations every day, and the number of active users was huge;

However, with the popularity of other social products such as WeChat, QQ's dominance has come to an end.

Although there are still a lot of QQ users, some once active users now spend less and less time using QQ, and some users even stop logging into QQ for half a year or a year.

These users can be defined as lost users, but some users may not log in to QQ for nearly a month or two months, but are still recharging membership for their QQ accounts. In this case, these users cannot be defined as lost users.

Therefore, when defining lost users, we need to determine the specific user categories based on the characteristics of the platform.

In order to quickly and accurately identify the target group of the product, companies can define lost users as: suspected lost users who have not performed any key operations over a long period of time; however, it is still necessary to combine the characteristics of the product to determine the key points.

For example:

1) The point at which users purchase products

The key points for users to purchase different products are different; for example, the key point in a music platform is to purchase music or its derivatives, while the key point in a shopping platform may be to browse or purchase goods.

2) Length of time without access

If a user does not visit the company's platform or purchase products for a month, this user can be considered a lost user, such as some community or dating platforms such as Weibo and Momo.

But for shopping platforms, it may take three months or even half a year before a user is identified as a lost user if it is found that the user has not visited the platform even once.

2. The second misunderstanding: mistakes in sample selection

This is because the company did not do a good job of prior investigation when selecting sample data before analysis, which resulted in the mixing of real return users and fake return users in the adopted data for analysis, resulting in some biased analysis results;

Therefore, before doing user churn analysis, you should first analyze the characteristics of returning users and exclude those users with longer active cycles.

It is relatively simple to calculate the total number of user churn on a platform. For example, if we consider users who have not logged in for more than one month to be churned users, then the total number of user churn is the number of users whose "current time point - the user's last login time point > one month" period.

However, calculating only the total number of churned users is not very meaningful for analyzing user churn, because in most cases, the calculated value is always increasing;

What companies need to do is to calculate the ratio of total number of lost users to the total number of users, as well as the number of new lost users, and observe their changing trends.

Whether a user is a lost user is determined based on the time of the user's most recent login. Therefore, to analyze lost users, it is necessary to find out the most recent login time of each user. The results of different websites should be different, because the time interval for each website will be different, and the longest may even be up to one year, which creates some obstacles for companies to obtain data.

Of course, in order to facilitate the analysis of registered users, companies can establish corresponding data tables in the database to store user information or choose professional data intelligence service providers such as Zhuge; while establishing basic user information, record the last login time of each user, so as to accurately calculate the time interval between the last login of each user and the current time, and use this to distinguish which users are lost users.

3. The third misunderstanding: not analyzing user behavior

Not analyzing user behavior means that when a company discovers user churn, it simply assumes that these users have churned, but does not know the reason for their churn, nor does it actively analyze the user's churn behavior, and thus does not understand the deeper reasons behind user behavior. To analyze user behavior, we must start with user stickiness, which includes the user's access frequency and the duration of the access interval.

1) Visit frequency: The visit frequency of users can reflect whether they are interested in the company's products and whether they have the urge to buy them. Some studies believe that when stickiness has not yet occurred, partial stickiness is equivalent to loyalty, so it can be considered that user loyalty is a prerequisite for user stickiness.

2) Visit interval: duration. If the platform does not spend time and energy to maintain users, or the product is no longer updated, then even if user stickiness was generated before, it will disappear over time. Users will not wait forever. Even if they have formed a habit of use, they will turn to competitors to buy substitutes due to demand.

For example, there is a restaurant where many people line up to buy breakfast every morning. Next to this restaurant is a small restaurant, but because it does not look high-end enough, not many people go there to eat.

Later, the restaurant, which was doing booming business, closed for two months because the owner had family matters, while the small restaurant next door remained open. When the restaurant owner came back, he found that the people who used to line up every day no longer came to his restaurant to buy breakfast, because the employees were used to eating at the small restaurant. Over time, the restaurant's business became worse and worse, and finally it had no choice but to close down.

The above example well illustrates the relationship between loyalty and user stickiness. At first, the employees liked to go to the restaurant for breakfast because the restaurant looked high-end, and customers formed a habit, which is loyalty.

So people would rather wait in line than eat here, this is user stickiness.

This kind of stickiness makes it easy to gain users, but also easy to lose users, because users are not indispensable to it;

Therefore, after the restaurant suspended business, customers went to small restaurants to dine in search of alternatives. Over time, users developed loyalty to the small restaurants, which led to a large number of user losses at the restaurant.

User churn analysis also needs to analyze the reasons for user churn from the product perspective in order to effectively control user churn fundamentally;

We need to classify the lost users, keep track of them, and optimize subsequent products based on the analysis results.

Understand users' needs, improve product functions in a timely manner, and transform and upgrade products at the appropriate time.

We should not only analyze the data of our own products, but also understand the real needs of users. We should know what users need and what kind of product functions are popular, so that we can further optimize and upgrade the products based on these data.

It is necessary to optimize products, platforms and other aspects from the user's perspective, such as setting the user's hot search page in a prominent position, adjusting the order of the product display page, optimizing the layout of goods, etc.

Solving the problem must start from the root cause. Only by improving product quality and optimizing product functions can the user churn rate be minimized.

Author: Zhuge io

Source: Zhugeio