Essential skills for product retention and growth

This article will share how to further find clues for retention growth through refined data analysis.

1. Product Loss Analysis

Before analyzing product retention, we need to first consider why users stay and why they churn? From which angles should we start to think about how to reduce churn and increase retention?

There are many reasons for user loss. Here are some common reasons for loss:

Insufficient perceived value of the product by users:

• Although the product has long-term value, its short-term value is unclear.
• The solutions provided by the product are not adequate and fail to meet the diverse needs of users.
• The product cannot continuously output new value to users.

Users believe that the product is difficult to use:

• There is insufficient guidance and incentives for users’ typical user paths, and user usage habits have not been developed.
• The user experience of the product is not good and users cannot get satisfaction from the product.

The cost of replacing products by users is low:

The product does not have an effective user incentive mechanism to make users invest effective time, money, manpower and other costs in the product, resulting in very little collateral losses when users change products.

2. Product Retention Difficulty Assessment

Before analyzing product retention, in addition to analyzing the reasons for churn, it is also necessary to assess the difficulty of retention based on product characteristics.

We can evaluate from the following seven aspects.

Regarding the above 7 aspects, we mainly focus on the following 7 questions:

1. How strong is the user's demand for the product, is it a painkiller or a vitamin?
2. How complex is the product's functional module? Is it a single function or multi-function and multi-module? Is it scalable?
3. What is the natural life cycle of the product? Is it long-term or short-lived?
4. What is the natural usage frequency of the product? Is it a high-frequency product or a low-frequency product?
5. Does the product have a large number of broadly substitutable products?
6. Is the cost for users to switch to other products high or low?
7. Is the product's monetization capability strong or weak?

We can use a retention difficulty assessment table to clearly judge the retention difficulty of the product.

3. Detailed analysis to find retention clues

There are many analysis ideas. Today I will mainly share two of them: user segmentation comparison and feature retention matrix.

3.1 User Group Comparison

Main thinking pattern: By comparing the retention curves of different user groups, observe users with different attributes and behavioral characteristics, whether the retention curves are different, and what are the differences.

The attributes and behavioral characteristics mentioned here can be considered and grouped from dimensions such as customer acquisition channels, user portraits, and user behaviors.

3.2 Functional Retention Matrix

Main thinking pattern: For multi-functional/module products, compare the retention rates of different functions and the proportion of active users using the function, observe the differences and find retention clues.

In addition, around the different functions of the product, other extensions and comparisons can also be made (presenting different functional retention matrices), such as:

• By comparing "new user usage" and "retention", we can observe the popularity of different features among new users.
• By comparing "function activity" and "days of use", we can observe the activity and participation levels of different functions.

3.3 User Lifecycle

In some cases, we can also improve retention through key indicators at various stages of the user life cycle, such as:

• Improve the activation rate of new users: As more users reach the exciting moment, the base of retained users will increase, which will move the starting point of the retention curve upward, thereby improving user retention overall.
• Improve new user retention rate: Improve the initial part of the retention curve, which will be passed on to the middle and late stages of the retention curve, thereby improving the retention rate.
• Improve the recall rate of new user churn: reduce the churn of later users and move the back end of the retention curve upward, thereby improving the retention rate.

3.4 User Engagement

Improving user engagement is also one of the important means to improve user retention. To increase the engagement of retained users, you need to increase the intensity and frequency of use:

• Intensity of use: Improving intensity can increase the value users get from each use of the product, thereby improving retention.
• Frequency of use: Increasing the frequency can consolidate and strengthen the user's habit of periodic product use, thereby improving retention.

The steps to analyze user engagement are shown in the following figure:

(1) Confirm product suitability

Engagement analysis is more suitable for high-frequency and high-engagement products such as social, content, and games. For some products, users can obtain value without high participation, such as low-frequency but high-unit-price second-hand transactions, SaaS products, etc.

(2) Calculating participation level

Calculate frequency:

1. Analyze the product's usage cycle, if it is months;
2. Draw a table showing the distribution of monthly active users by number of active days per month;
3. Group monthly active users by frequency.

Computational Intensity:

1. Analyze the product's usage cycle, if it is months;
2. Draw a table showing the distribution of monthly active users by usage time;
3. Group monthly active users by different durations.

(3) Establish ideal indicators

Set an ideal frequency and intensity target based on the user's natural needs and usage habits for this type of product, as well as current data.

(4) Further analysis to find clues

1. Compare and analyze the usage frequency and intensity of different user groups to find clues;
2. Compare and analyze the frequency and intensity of use of different product features to find clues;
3. For a certain function, find out the factors that hinder users from increasing the frequency and intensity of use.

IV. Reference Cases

In this section, the editor uses a simple case to introduce the specific operational process of how to discover some growth clues through statistical retention data. It mainly shares ideas and implementation processes, and the specific data content is not authentic.

Through the link, we can see a group retention data (virtual data, no confidential data involved). Suppose this is a user group retention data table for an audio product.

4.1 User Profile of Chinese Online Audio Users

Since we want to analyze user purposes, we must first have a certain understanding of the users, so here the editor will briefly introduce the basic information of Chinese online audio users.

The data shows that the majority of online audio users are office workers born in the 1980s and 1990s, among which those born in the 1990s account for 52.5%. According to the research conclusion of iMedia Research: the active time of online audio users is mainly concentrated in the evening and noon, among which the evening (18:00-23:00) accounts for 40.8%; in terms of usage scenarios, meal breaks (including lunch and dinner) and before bed are the two main scenarios.

Judging from the distribution of reasons for use in the above figure, online audio users use online audio mainly to achieve the purposes of "relaxing the body and mind" and "leisure and entertainment", followed closely by "relieve emotions" and "kill time", which are more in line with the usage scenarios, indicating that online audio users have a large demand for entertainment.

According to the above figure, iMedia Research concluded that music, audiobooks and news information are currently the most popular audio types.

4.2 Product Retention Difficulty Analysis

First, we conduct a simple retention difficulty assessment for the product itself:

• User demand: Users’ demand for online audio products is not a rigid demand, but after the products have cultivated good user habits, the demand intensity has increased.
• Product functions: In addition to recording and listening to audio, there are no particularly complex and diverse product functions, and the business complexity is mainly concentrated in the content matrix.
• Life cycle: The life cycle of an audio product varies greatly depending on the richness of the content matrix. Taking all factors into consideration, the life cycle for listeners is medium.
• Frequency of use: According to statistics, online audio products are high-frequency products.
• Substitutes: Although the online audio industry has experienced brutal baptism and product elimination, there are still many homogeneous products.
• Replacement cost: The product's functions and technologies are not irreplaceable, so the replacement cost is low.
• Monetization capability: As a content consumption product, its monetization capability is not strong, far inferior to online video products.

Therefore, the following conclusions are drawn:

Comprehensive average score: 5 points.

Product retention difficulty: medium to high.

4.3 Analyze data based on user retention curve

The following retention curve and channel download distribution chart are obtained from the original data:

Figure 2 (Note: the red line is the average value)

The following data phenomena can be obtained from the retention curve:

Recording audio: The overall retention data performed the best. Although it was slowly declining, the retention rate remained stable at above 60%. Downloads accounted for the smallest proportion, only 7.73%.

Children listening to stories: The overall retention data performance is not as good as "recorded audio", ranking second, and the retention rate is steadily declining, and it will remain above 55%. , the download volume share ranked second from the bottom, at 12.04%.

Current Affairs News: The overall retention data performance is poor.

Learning and recharging: The overall retention data performance is very poor, and the overall retention is not higher than the average. And the download volume accounts for only 15.7%.

Leisure and entertainment: The overall retention data performed the worst, with the first-week retention rate being only 80%, 5 percentage points lower than the average. And the first-month retention rate dropped significantly. It eventually fell below 40%. Downloads accounted for the highest proportion, at 40.31%.

4.4 User Purpose Retention Matrix

Use the function retention matrix to conduct detailed data analysis and convert the data table into a user purpose retention matrix diagram, as shown below:

Vertical: monthly retention, horizontal: monthly active user ratio

From the above figure, we can find that the retention of user usage purposes is obviously polarized. Combined with user portraits, we can draw the following conclusions:

Recording Audio:

• Trend phenomenon: "Recording audio" has high retention and low monthly active share;
• Optimization strategy: "Recording audio" is a rigid and low-frequency operation in audio products, and the user base is relatively small. The strategy to increase active users should focus on improving the friendliness and convenience of the interactive experience of the "Recording audio" function.

Children listen to stories:

• Trend phenomenon: "Recording audio" has high retention and low monthly active share;
• Optimization strategy: Although the user scenario of "Children Listening to Stories" accounts for a relatively small proportion in the "2019 Online Audio User Usage Scenarios", with the huge children's consumer market brought about by the two-child policy, we should focus on improving the richness and fun of the content matrix of this section.

Current News:

• Trend phenomenon: "Recording audio" has high retention and low monthly active share;
• Optimization strategy: "Current Affairs News" accounts for a high proportion of "Audio Type Preferences of Online Audio Users in 2019", reaching 43.5%. However, the user proportion of this product is relatively low, with 11,263 downloads in half a year, which is far from the average value of 14,347. The exposure positions and push strategies should be increased more reasonably.

Learning and charging:

• Trend phenomenon: "Learning and recharging" has low retention and high monthly active share;
• Industry analysis: In "Reasons for online audio users in 2019", knowledge learning accounted for 12%, and in "Audio type preferences of online audio users in 2019", knowledge courses accounted for 22.5%, both of which are at a lower-middle level. The active time of online audio users is mainly concentrated in the evening and noon. In terms of usage scenarios, eating and resting (including lunch and dinner) and before going to bed are the two main scenarios. Therefore, the proportion of monthly active users with the goal of "learning and recharging" is close to the median value, which is an objective reason that cannot be ignored;
• Optimization strategy: Because the behavior of "learning and recharging" is mostly continuous, and the retention rate of this type of behavior of this product is low, it may be that the quality of the learning audio content is low, or it is a single video, and there are few series of learning audios. It can improve the content quality of learning audio. Introducing a series of high-quality audio content.

Leisure and Recreation:

Trend phenomenon: "Leisure and entertainment" has low retention and high monthly active share;

Industry analysis: Industry data shows that the audio type preferences and reasons for use of "leisure and entertainment" users account for the largest proportion. Then this product has the lowest retention in this regard, but the highest monthly active users, which is twice the average. This indicates that the product has sufficient exposure to the "leisure and entertainment" module, but the content cannot attract user retention. It may be that the content type does not match the user group, or the content type is not consistent and continuous, or the content matrix is ​​not rich and diverse enough.

Optimization strategy:

1. If the content type does not match the user population, then we need to make a judgment based on data such as user gender, age distribution, and location distribution. For example, male users prefer finance, novels, history, etc., while female users prefer literature, romance dramas, and emotional genres.
2. If the content type is not coherent and continuous, then the proportion of audiobooks or series audios should be increased. This is because individual audios are independent of each other and the closeness between audios is not as strong as that of audiobooks, so the continuity and time of users' listening will not be too long.
3. For example, the content matrix is ​​not rich, diverse, or interesting enough. Then it is necessary to increase UGC products, purchase some high-profile IPs, and guide them with KOLs.

Author: Yang Sanji

Source: Yang Sanji (zyjn2020)