Essential skills for increasing user acquisition and retention

In product operations, user retention rate is one of the key indicators that everyone pays most attention to. So how can we more efficiently find some clues about user growth from user retention and formulate corresponding growth strategies?

1. Overall perception of user retention and user activity

1.1 The underlying core of retention and user activity

1. Regarding retention and activation, its essence is to find the most cost-effective conversion and growth path for different types of users as the user base grows, and then provide guidance and incentives.
2. The establishment and stability of the long-term value of the product is the premise for retention + activation to be meaningful;
3. Products with different business types and attributes are bound to have different focuses when thinking about "retention";
4. The richer the product form + functions, and the more user behaviors, the higher the technical content of "retention and activation";
5. After the typical user conversion and growth paths are clear and the closed loop of user usage habits is formed, a user incentive system can significantly amplify the efficiency of internal growth through productization.

1.2 Growth/Retention Work Methodology

Most people think of the growth/retention workflow as follows:

But the actual growth/retention workflow should be as follows:

• All “growth” work ultimately involves building typical user paths and guiding and motivating users to achieve conversions.
• The basic premise of growth/retention/activation work is that you have a "clear and driven user conversion/growth path."
• The purpose of all our work on a product is either to discover it or to amplify it. And this growth path may also change constantly.

2. Analyze the problem through user retention data

2.1 Definition of Retention Curve

When it comes to the retention curve, we need to first explain what retention rate and churn rate are.

Extending from the above definition, the retention curve is: continuously tracking the changing trend of retention rate of user groups that became active at different times over time.

2.2 How to draw a retention curve

2.2.1 Determine the core ideas of key behaviors

• Initial behavior: Which initial behavior must a user complete to be defined as a retained user?
• Return behavior: When a user visits the product again, what behavior does he need to complete in order to become a retained user?

Then the initial behavior and the return behavior are the same in most cases. Of course, there are differences in products in certain specific scenarios. The key is determined by the needs themselves.

2.2.2 Methods for selecting time periods

The natural life cycle of different products is different, such as:

• Investment products: weekly~monthly;
• Social products: every day;
• Game products: every day;
• Content products: daily to weekly;
• Insurance products: weekly~monthly;

When you don’t know the natural life cycle of a product, how do you go about finding this time period?

Method: Analyze the distribution of monthly active users’ active days in a month;

• Use monthly active users as the denominator;
• Further divide the proportion of users with different active days in a month;
• Find the active days item with the largest user proportion, which is the natural usage cycle of users in this retention curve.

2.2.3 Collect data and create forms

• Record the number of users who complete the initial behavior for the first time in each cycle, which is generally the number of activated users;
• Track the number of users who continue to return in each subsequent cycle. Generally, this is the number of users who have retained key behaviors.
• Through 1 and 2, calculate the percentage of users who return in each cycle to the number of activated users in the first cycle.

Through the above operations, you can get the retention data form provided in case 2.4.

2.3 How to find retention problems from the retention curve

2.3.1 Basic form of evaluation curve

1. Downward trend

Retention curve: indicates that PMF has not been achieved.

Focus on changing your product to find a value proposition for your core user base, and then expand on that base. Don’t start blindly attracting new customers at this time, otherwise you will just be “a tree without roots”.

2. Flattening

Retention curve: indicates that the product has reached PMF.

This indicates that among users who have tried the product, a certain proportion have discovered its value and continue to use it after a period of time, and it is time to start attracting new users.

3. Smiling type

Retention curve: Not only can PMF be achieved, but there are also a large number of returning users.

The most ideal retention curve generally only exists in specific types of products.

2.3.2 Compare the industry average and observe the changing trend

When we analyze the user retention of a product, we must also consider the overall trend of the industry to which the product belongs. Compare data differences and product differences with leading products.

Thinking patterns for observing changing trends:

• Compare based on the time dimension;
• Compare the before and after performance of your growth strategies in real time.

Compare the new retention curve with the old retention curve to observe whether the new retention trend has a higher starting point, a smaller slope, and a smoother curve than the old retention curve.

2.4 Case Analysis

In this section, the editor uses a simple case to introduce how to discover some growth clues through statistical retention data.

Through the link (https://shimo.im/sheets/K8dqhKtQ8GJPw38P/MODOC/), we can see a weekly retention data (virtual data, no confidential data involved). Suppose this is a user weekly retention data table for an audio product.

First, assume that the statistical premise of the data is as follows:

• Retention behavior: first login
• Time period: Week
• Return visit behavior: Log in again
• Retention rate: How many users who log in for the first time log in again within the next week.

Because this is a weekly retention data table, the data needs to be visualized before analysis. The editor is lazy and directly uses Excel for data visualization.

First, the average retention curve is drawn as follows:

From the above figure we can find the following trends:

• The first-week retention rate is only 85%, which means that 15% of users never log in again after the first download.
• After the second retention rate in the first month dropped rapidly to 61%, the rate of decline slowed down significantly and tended to decline steadily;
• From the 3rd week to the 17th week, the data remained basically between 50% and 60%;
• From week 18 to week 36, the data dropped again, remaining stable between 39% and 43%;
• From week 37 to week 45, the data first experienced a sharp drop of 2 percentage points, and then fluctuated and fell to 35%;
• After the 46th week, the data fell further and finally dropped to 23%;
• In summary, this retention curve is a declining retention curve;

Based on the above phenomena, we further process the retention data table in the case and obtain the retention heat distribution diagram as follows:

The above conclusion is further verified through the weekly retention chart (red represents good, green represents poor).

After zooming in on the above data heat map, we can find several data anomalies, which may also be clues to growth:

• From the download volume table, it can be clearly seen that there is a significant gap between the download volume in the four time periods of "2018/4/22", "2018/9/23", "2018/11/18" and "2018/12/23" and the download volume in the previous and subsequent time periods.
• On "2018/4/22", i.e. the 16th week, the overall data indicators were nearly 10 percentage points lower than the previous and subsequent data, with obvious abnormalities;
• From the heat map, it can be found that starting from the 46th week, the data may become abnormal and show a significant decline. In order to verify the conjecture, the horizontal heat map (month-on-month comparison) is changed to a vertical heat map (year-on-year comparison), as shown in the following figure:

The data trend once again confirms that the above speculation is correct.

Based on the above data analysis, the following hypothesis is proposed:

1. The average retention rate in the first week was only 85%, and 15% of users were lost after the first download. It may be the result of inaccurate channel promotion, or it may be that users did not reach the exciting moment after their first login, resulting in rapid loss.
2. On "2018/4/22", the 16th week, the number of downloads surged, increasing by 100% year-on-year. This may be due to increased exposure in new channels and follow-up operational activities. However, the retention rate has dropped sharply. It is speculated that the effect of the new activity has not met the expected effect of users. The retention rate brought by the new operation activity is much lower than the retention rate brought by the original activation process, which has lowered the overall market data.
3. On "2018/9/23", the number of downloads reached the peak of the whole year, and the weekly retention rate also reached the highest level of the whole year. It is speculated that the new user incentive limited-time activities may be added, and the channel exposure will be strengthened. From the year-on-year heat map, it can be found that the overall retention data has a significant short-term improvement after "2018/9/23".
4. The number of downloads has dropped drastically since November 18, 2018, which has directly affected the user retention trend throughout the year. It is speculated that there may be a major version optimization and a transformation in the overall product positioning, but the effect will not be ideal.

Combined with the user life cycle, the following growth clues are obtained:

New user activation phase:

• North Star Metric: New User Activation Rate
• Judgment basis: Among the strategies for improving user retention through the life cycle, improving the activation rate of new users has the highest priority because it has the greatest impact and the lowest operational difficulty.
• Data phenomenon: The average retention rate in the first week is only 85%.
• Growth strategy: Improve the accuracy of channel promotion and increase the proportion of new users who reach the activation moment.
• Implementation plan: Through refined data analysis of the percentage of retained users and active users in different channels, we can find optimization directions and improve the accuracy and conversion rate of channel delivery.
• Optimize the new user activation process and increase the proportion of new users who reach the activation moment.

New user retention stage:

• North Star Metric: First-Month Retention Rate of New Users
• Judgment basis: Among the strategies for improving user retention through the life cycle, improving the retention rate of new users ranks second in priority because although the impact is similar to improving the activation rate of new users, the operation difficulty is relatively high.
• Data phenomenon: The first-month retention rate of new users is only 63%. There is a lot of room for data growth.
• Growth strategy: Increase the proportion of new users who reach activation time and increase the activation rate of new users
• Implementation plan: Build user portraits and analyze the purpose and behavior of users in using the product under different user dimensions. Through detailed data analysis of the retention rate and active user rate of users with different usage purposes, we can find optimization directions and improve new user activation strategies for users in different groups.

Lost user recall stage:

• North Star Metric: New User Recall Rate
• Data phenomenon: Starting from the 36th week, the retention rate dropped significantly, from 39% to 23%.
• Judgment basis: Among the strategies for improving user retention through the life cycle, improving the recall rate of lost users ranks third in priority because the impact is not as good as the first two, and the operation is not easy.
• Growth strategy: Recall users who agree with the product positioning and have demand for the product but have already left.
• Implementation plan: Push marketing activities through SMS push and other actions;
• Further improve the precise delivery of high-quality content (slot display, Push push, etc.);

This is all I have to say about basic retention data analysis.

Author: Yang Sanji

Source: Yang Sanji (zyjn2020)