How to prevent user churn?

The main tasks of user operations include: user acquisition, user activation, user retention, user payment, and user dissemination, which are the main contents of the AARRR model.

However, the AARRR model focuses more on guiding user behavior and improving user value step by step from the moment the user comes into contact with the product, but does not consider the preventive operations of user churn. When users no longer use the product, any value enhancement work is useless, and the high customer acquisition cost is difficult to recover as users churn.

Therefore, user churn prevention operations play an extremely important role in user operations.

This time, I will share with you the operational ideas and methods for user churn . Don’t say goodbye to users easily!

1. Product Lifecycle: When to Care About Churn

All products should focus on churn, but the priorities and investments vary.

The product life cycle theory divides products into four stages: introduction, growth, maturity, and decline. Different stages have different user operation focuses. In the early stages of the product, the main focus is on user acquisition and retention, while attention and investment in user churn are mainly in the maturity stage.

Products in the mature stage have a large number of users, are relatively mature and have a stable experience, and have accumulated a certain amount of user data, which provides better support for user operations. Investing in operations to prevent user churn at this stage will make it easier to achieve positive input-output. Of course, if there are sufficient operational resources, it is still recommended to carry out preventive operations to prevent user churn as early as possible.

2. User churn type: Which users are worth operating?

Before conducting user churn prevention operations, we need to understand why users churn so that we can design targeted churn prevention strategies. Based on the reasons for user churn, I divide user churn into four types:

1. Natural loss

Natural churn occurs when user demand/interest wanes or disappears, and they no longer need to use our product.

The user loss of driving test products is mainly natural loss. Users start using the product because they need to obtain a driver's license, but after obtaining the driver's license, they no longer need the product, so natural loss occurs. In addition, once popular products such as Footprints and Face Meng attracted a large number of users due to their novel and creative product functions, but they only satisfied users' short-term interests. After user interest waned, a large number of users left.

It is difficult to operate to deal with natural attrition through methods such as pre-attrition prevention and post-attrition recall. Instead, it must be solved from the product and service level.

One idea is to enrich product and service content and extend the user life cycle. Nowadays, driving test products have many modules. In addition to driving test services and content, car buying services, car owner communities, etc. have also been added. After users complete the driving test, there are still corresponding functional services to meet their subsequent needs. If the product positioning is more vertical, it is difficult to break through user cognition. The effect of this idea is poor. It is difficult for users to get used to buying cars and communicating through driving test products. Moreover, products such as Autohome and Dongchedi have already occupied the minds of users.

Another idea is to build a product and service matrix to take over naturally lost users. By using multiple products and services to meet the different needs of the same users at different stages, the independent positioning of each product also makes it easier to acquire and retain users. The leading companies in the online education industry have all developed preschool enlightenment products and K12 stage products. K12 stage products can take over users who no longer need enlightenment products as they grow older, thereby maximizing user value.

2. Flexible loss

Flexible churn means that users still recognize the product, but churn due to subjective reasons.

Taking fitness products as an example, users start using them because of needs such as weight loss and bodybuilding, but due to user laziness it is difficult for them to stick to the product and they eventually leave. Users still recognize the product, still envy the fit body, and still have the need to lose weight and exercise, but they subjectively choose to give up. There are also some low-frequency or niche products that users use less frequently, but they are still relatively recognized. However, as time goes by and due to busy lives, users gradually leave.

For flexible churn, targeted recall outreach and guidance incentives are better ways to deal with it . Excellent products are meant to help users become better.

3. Rigid loss

Rigid churn means that users still recognize the product, but churn occurs due to objective reasons.

The objective reasons leading to user churn can be divided into user-related reasons and external environmental reasons.

Let’s first talk about the user’s own reasons, taking financial products as an example. A user has been purchasing funds and other financial products on a certain financial product, but he suddenly encountered a financial crisis and withdrew all the money in the financial product. He also could not continue to use the financial product because he had no surplus funds in the short term.

This is rigid loss caused by the users themselves, but in fact the users still recognize the product and are unable to continue using it due to objective reasons. This kind of loss is not real loss. When users have surplus funds, they will most likely continue to use the product. Periodic recall can reduce the actual churn of such users.

As for external environmental reasons, they are mainly changes in social environment and market policies. During the epidemic, cinemas were closed and performances were not allowed to be held. Many ticketing products experienced a certain loss of users, but this was not a real loss due to objective reasons. Timely guidance and recall can retain such users.

Although rigid churn is churn, it is not churn in the true sense. The input-output ratio of user recall is the highest, so such users must not be ignored.

4. Experience loss

Experience churn refers to the churn caused by the product not meeting user needs well and the experience not meeting user expectations.

There are two situations of experience churn. One is that the product function is insufficient and does not meet user needs well, so users give up using the product and look for alternative products. In this case, we need to receive user feedback to maintain product iteration, and then cooperate with churn recall to handle it;

Another situation is that there is no problem with the product function, but users do not perceive the product value well. There are problems with the product process and experience. In this case, optimizing the usage process and actively guiding users to better use the required functional services can reduce churn.

3. Churn operation mechanism: prevention beforehand and recall afterwards

To deal with user churn, it is necessary to prevent it before it happens and recall it after it happens. Let’s take a closer look.

1. Loss prevention mechanism

After users have been silent for a long time or even completely lost, the input-output ratio of intervention will be very low. The correct response is to establish a complete loss warning mechanism, identify users at risk of loss in a timely manner, intervene and guide, and maximize user retention.

The loss prevention mechanism can be built in four steps:

1) Define lost users: What kind of users are lost users?

2) Analyze the signs of churn: characteristics of churned users and their behaviors before churn

3) Establish an early warning mechanism: monitor data and identify potential lost users

4) Conduct intervention and guidance: Intervene based on user characteristics and reasons for churn

To define lost users, we need to follow a basic framework, namely "how long (time) the user has not performed a specific behavior (action)", including two main conditions: key behavior and time length. The key behavior is generally login access, and some products also choose payment, publishing, etc. as key behaviors.

The judgment of the length of time requires specific data analysis based on product and user characteristics. One idea is to analyze the average user visit cycle (how often a user uses the product) and the length of time a silent user returns to the product (how long it takes for a long-term inactive user to use the product again).

Analyzing the signs of churn is a prerequisite for identifying churned users in advance.

On the one hand, characteristic hypotheses can be made and verified based on the four types of user churn, and user survey interviews can be conducted when necessary. On the other hand, churned users can be directly identified, and their behaviors before churn can be specifically analyzed. churned users can also be stratified and analyzed based on dimensions such as channel, region, and age.

Establishing an early warning mechanism is more about managing user churn through productization. Based on the definition of lost users and signs of loss, using behavioral tracking and data quantification as identification methods, we can promptly find and identify users at risk of loss and conduct timely monitoring.

Intervention guidance means timely intervention after identifying users at risk of churn. Carry out targeted intervention based on user characteristics and reasons for churn. Common intervention methods include information reach, preferential guidance, activity rewards , etc., so that users can use the product again in a timely manner instead of directly churn.

At present, the churn intervention of major products has become relatively mature. If you pay more attention to the recall notifications of Taobao, Didi, etc., you will be able to discover the basic intervention methods for users at risk of churn.

2. Lost customer recall mechanism

After the churn warning mechanism has been implemented, users still gradually give up using the product and become real churned users. At this time, the input-output ratio of recalling churned users is low, but users are our assets and we cannot give up easily.

Conventional reach-out recall methods that use SMS/push as the main form and daily activities/promotions as the main content are no longer of much value for completely lost users. These operational methods have been used when users are identified as at-risk of losing. For completely lost users, more personalized and targeted recall methods are needed.

By stratifying the characteristics of lost users and conducting data analysis, we can take targeted measures to recall them. However, the frequency of recall cannot be too high. On the one hand, disturbing users excessively will not be worth the cost. On the other hand, recall text messages generally cost 0.03 yuan per message, which is a cost.

Users who have naturally lost often need to be recalled during product feature iterations and major events to meet their needs again.

Users who churn flexibly need to be recalled in a relatively high-frequency and periodic manner, and their service ideas need to be changed to provide incentives and guidance.

Rigid churn users still have clear needs for the product, but are temporarily unable to use it due to objective reasons. Periodic caring visits and proper contact with users are needed to ensure that users are more likely to return and use the product when the objective restrictions disappear.

Users who have experienced churn need to be recalled after optimizing the product experience.

Many times, it is difficult to accurately classify user churn types, so general recall methods must also exist, but the content and frequency must be controlled. When user recall is not effective, one-on-one interviews with some of the churned users may yield new findings that differ from data analysis and churn assumptions.

4. Churn operation case: e-commerce churn operation mechanism

In order to help everyone better understand and apply the above theories and methods, let’s take the user churn prevention operation work of maternal and infant e-commerce products that I was once responsible for as an example to see how to plan and implement the user churn prevention operation mechanism. (Some data are not accurate and are only used for case illustration)

Let me first introduce the basic situation of this maternal and infant e-commerce product. The platform's products are mainly maternal and infant products, and have gradually expanded to the field of household consumer goods (daily necessities, beauty and clothing, digital appliances, etc.). Maternal and infant products contribute more than 75% of the sales of the entire platform. In terms of product users, more than 70% are female users aged 25-45, and there is a platform membership system. The consumption amount of member users accounts for more than 60% of the total sales of the site. The platform has been established for more than 5 years, with a total of tens of millions of registered users, and has entered a relatively stable development stage.

It can be seen that the product and user characteristics of this maternal and infant e-commerce platform are relatively obvious, and the product stage is mature. The user growth rate is gradually slowing down, and user loss due to competing products is increasing, so preventive operations to prevent user loss are very necessary. Following the user churn prevention operational steps introduced in the previous article, we will introduce our team’s operational actions at the time.

1. Define lost users

The definition framework for churned users is "how long has it been since a user performed a specific behavior?" When defining churned users, we distinguish between member users and ordinary users because the activity characteristics of member users and ordinary users on the platform are quite different.

Ordinary users visit the platform once every 10 days on average, while member users visit the platform once every 5 days. Member users are obviously more active. Among silent users (those who have not visited the platform for 30 days) who return again, member users have a longer return cycle, which also shows that it is easier to recall member users and their retention is better. Based on the average visit cycle and the revisit cycle of silent users, as well as the analysis of user characteristics and value, we determined the churn definition for two types of users, and formulated a stricter churn definition for highly active member users.

2. Analyze the signs of churn

Next, we conducted further analysis on member users and ordinary users, mainly analyzing the characteristics of the two types of churned users and their pre-churn behaviors. The characteristics considered included user source channels, user shopping preferences, user consumption amounts, etc. The pre-churn behaviors mainly focused on visits and shopping.

We found that although churned users visited the platform many times before churn, their shopping behavior decreased significantly. Therefore, we combined the long-term non-shopping characteristics of users with the churn cycle to determine the main signs of user churn:

1) Signs of general user churn: not visiting the platform for 20 days and not making any purchases in the previous three visits;

2) Signs of member user loss: No visit to the platform for 14 days and no purchases in the previous two visits.

According to the platform's historical data, only 15% of users who met the churn signs criteria did not churn and remained, while the rest of the eligible users all churned. Moreover, reverse verification based on the data of lost users shows that more than 90% of lost users meet the characteristics of loss signs.

3. Establish an early warning mechanism

After clarifying the definition of lost users and the signs of user loss, the next step is to establish an early warning mechanism. The company purchased and applied a third-party user management system to assist in user operations, while large companies developed their own.

We directly define user groups in the user management system based on user attributes and behavioral data, mainly screening user access data and shopping data. When users meet the churn sign conditions and churn user conditions, they are automatically classified as churn risk users and churn users respectively. User operations will monitor the data of churn risk users and churn users on a daily basis and provide intervention and guidance.

4. Conduct intervention and guidance

The user management system enables us to monitor users at risk of churn and churned users at any time. When a user is classified as a churn risk user, we begin intervention and guidance, and cooperate with system functions and manual operations to reach and recall the user. We conducted a total of three reach and recall actions, with different forms and contents, as follows.

The three contacts for users at risk of churn are progressive and the cost is gradually increased. Only users who have not returned after each contact will enter the next stage of contact. If the user returns after being contacted, he/she will enter a separate return user pool for monitoring to reduce further churn.

The above three steps are only the main reach and recall framework. In actual implementation, personalized reach and recall will be carried out based on user characteristics.

For example, some mother users mainly consume maternal and infant products. As their children grow older, they no longer need milk powder and diapers, and natural loss occurs. At this time, we push targeted products and activities such as clothing, beauty, nutrition and health care, and the copywriting is also designed, such as "Take good care of your children and love yourself", so that these naturally lost users can find that the platform can not only provide maternal and infant products, but also high-quality and affordable household products, thus achieving retention.

For example, when reaching out to high-risk users who prefer beauty products, beauty coupons can be pushed to them; when reaching out to users with low historical spending amounts, low-priced products with free shipping can be pushed to them. These are all targeted ways of reaching out and recalling users, which can further enhance the effect of reaching out and recalling high-risk users.

After calling the user back several times, the user remained silent and eventually left, so we did not continue to bother the user. On the one hand, during large-scale coupon events such as Double Eleven and anniversary days, we will uniformly recall lost users; on the other hand, we conduct one-on-one interviews and questionnaire surveys with some lost users to understand user feedback and then iterate products and operations.

In this way, a relatively complete user churn prevention operation mechanism can be put into operation. After nearly half a year of optimization and adjustment after implementation, the churn rate of users at risk of churn has been reduced by 50%, and the revisit rate of churned users has also increased by 20%. These users contribute millions of sales every month.

The above is my sharing on user churn and its prevention and response. I hope it can provide some reference and inspiration for your work.

You may have seen this sentence: " The cost of retaining an old user is much lower than attracting a new user ." Although it is impossible to verify, it is indeed a principle that every operator should keep in mind. Retaining users who once trusted and loved the product, and ensuring that users who leave the product are not too dissatisfied, is a task that we cannot ignore.

User churn is closely related to user retention. Doing a good job of retention means reducing churn.

Author: Wu Yijiu

Source: Wu Yijiu