Analyze user churn from 4 perspectives!

The author of this article, through his own project review, has sorted out for us from six aspects how to increase user growth and thus improve user retention rate through product iteration. I hope it will be helpful to you.

Let’s first use the STAR rule to give some background information:

S: A content product that has been online for nearly 2 years. At that time, the promotion funds were drastically cut, resulting in a significant decline in daily active users.

T: Complete product OKR goal - increase daily active users to X.

A: Through growth hacking methods, rapid iteration and trial and error, we find growth strategies in experiments and achieve user growth.

R: The APP retention rate before the revision was slightly lower than that of APPs of the same scale in the same industry; after the revision was launched, it maintained a good level for 5 months, and the APP second-week retention rate was about 20% higher than that of APPs of the same scale in the same industry.

Note: Due to commercial reasons, the data in this article has been anonymized and is for reference only.

The background is complete, and the following text begins - Review: How to increase user growth through product iteration and increase retention rate by 20%+

Growth thinking: find the direction of growth → determine the North Star indicator → break down the goals → conduct experiments → analyze and summarize the results → self-reflection

Step 1: Find the direction of growth

Growth direction → fission

Fission refers to the viral communication methods such as group buying, support, distribution, and bargaining that we often see in the circle of friends. Fission is only a part of the entire growth system, and it mainly appears in the customer acquisition (Acquisition) link.

Fission brings growth, but it is far from the whole of growth. Growth also requires attention to the entire life cycle of users. If the needs of users brought by fission activities cannot be solved through products, they will only be a bunch of passerby data, and no matter how good the operation is, it will be useless.

Growth direction → Increase the number of new users and active users

We often see some product press releases "XX daily active users exceed XX", "XX number of users reaches XX", "XX download ranking is X", and it seems that growth means the number of users/daily active users. These two data are undoubtedly very important and are often used as product KPIs.

However, the increase in DNU (Daily New Users) and DAU (Daily Active User) does not mean that customers are growing effectively.

The increase in new users and daily active users may be short-lived. Look at Luo Yonghao’s "Chatbao", which relied on the KOL halo and user sentiment to top the APP Store social list for 13 consecutive days and attracted nearly 8 million users within a month, but now the situation is bleak.

Compared with Luo Yonghao’s “Chatbot”, more products spend a lot of money on promotion to acquire users. In recent years, the price of channel promotion has become higher and higher, and the pricing of CPC/CPT/CPD/CPA has continued to rise. The CPT price of the app store has doubled in half a year. The DNU and DAU of users brought in by spending money are rising steadily, but perhaps the churn rate of the product itself remains high. In fact, customer retention is gradually decreasing. If users cannot stay in the product, this "growth" is meaningless.

Growth direction → Improve retention rate

Imagine a wooden barrel with several holes. The more water you pour into it, the more water will flow out. It is like a product attracting many new users but being unable to retain them. This will not lead to sustainable growth of the product. In addition, the mobile APP market is now becoming saturated, the bonus era is over, and acquiring customers is not as easy or cheap as before. We have now moved from an "incremental" era to an "stock" era, and it is more cost-effective to retain old users.

Zhugeio wrote in "Eight Data Analysis Models": For users, the higher the retention rate, the better the product grasps the core needs of users, and the stronger the dependence of users on the product; for products, the higher the retention rate, the more active users the product has, the greater the proportion of users converted into loyal users, and the more conducive to improving the commercial capabilities of the product.

To sum up, considering the current situation of the product, increasing user growth and improving "user retention rate" are more important directions to focus on.

Step 2: Determine the North Star Metric

Definition of the North Star Metric

The North Star Metric, also called the One metric that matters (OMTM), is the most critical indicator of the product at this stage. Products in different tracks have different North Star indicators for growth, and we need to find the only key indicator.

Choosing a North Star Indicator

Generally speaking, a good North Star indicator has the following properties:

• It is a metric that can be clearly measured (in this case, it is measuring "retention rate").
• It is an indicator that can reflect product value and customer value.
• It is a leading indicator, not a lagging indicator.

The selection of the North Star indicator is very important. If it is not selected properly, it may cause a series of subsequent wrong decisions by the team. Therefore, before determining the North Star indicator, we need to do two basic tasks - data analysis & user research.

Phase 1: Data Analysis

1. Content data analysis

The APP is a content-based product with four main business modules: "Homepage (i.e. article information)", "Q&A", "Report" and "Data". To study the interaction between people and content, we retrieved data from each section, such as the dwell time and number of launches of each section.

In addition, each section has its own unique data indicators, such as the number of readings, number of independent users, average reading progress, approximate completion rate, bounce rate, number of shares, number of favorites, number of likes, number of comments, etc. for the "Homepage"; the number of questions, question rate, number of answers, answer rate, etc. for the "Q&A"; the number of visits and downloads for the "Data", etc.

2. Focus on data analysis

The user's attention behavior is equivalent to telling you in black and white what he/she likes to watch. The relevant data is worth analyzing. Currently, the objects that can be followed in the APP are "Author", "Column", "Industry Questions" and "Reporting Agency".

3. User life cycle analysis

Here I did not divide users into "new users", "active users" and "returning users", but "loyal users" and "non-loyal users" because data analysis tends to be noisy and difficult to clean, and the greater the difference in user attributes, the faster a breakthrough can be found.

4. Output conclusion

Found that the trends of the following 6 indicators are positively correlated with "retention rate"

• Metric: Number of articles
• Metric: Number of followers of the author
• Indicator: Number of reporting organizations concerned
• Indicator: Number of database members
• Indicator: Number of column followers
• Metric: Report download rate

Compared with "non-loyal users", the number of followers of "loyal users" is much higher than that of the latter. The average number of followers of "loyal users" is ≥ 6 (Magic Number), and the type of followers they follow tends to be "Author" > "Reporting Agency" > "Column" > "Problem".

Phase 2: User Research

We invite APP users to conduct user research and understand users through user observation, in-depth interviews, focus groups, quantitative questionnaire surveys, etc.

The following are the labels that users give to our APP. Users use the APP mainly to obtain industry-related content.

In the question "Which of the following new features/experiences will increase the frequency of your opening the app?":

72.24% of the respondents thought that: the addition of the "Follow Industry" function would help them quickly find the content they want to watch;

70.63% of the respondents believe that: Set the push switch for "Contents You Follow" and enable notifications when the pushed content is updated;

63.52% of the respondents believed that: optimize industry classification and add more subdivided/hot/emerging industries;

You can see three keywords: "Follow", "Push" and "Industry".

Phase 3: Output Conclusion

Combining the analysis of the first and second phases with the specific business, I set the North Star Indicator as "Attention-Daily Active".

The "Follow-Daily Activity" formula is broken down as follows:

"Focus" = "Industry_Focus" + "Author_Focus" + "Reporting Agency_Focus" + "Column_Focus" + "Issue_Focus" (these 5 items are referred to as A/B/C/D/E respectively)

"Daily active users" = "Daily new users" + "Number of users retained the next day from yesterday's new users" + "∑N-day returning users in the previous N days (N>=1)" (these three items are referred to as F/G/H respectively)

Therefore, "Follow_Daily Activity" = AF+AG+AH+BF+BG+BH+CF+CG+CH+DF+DG+DH+EF+EG+EH

Step 3: Disassemble the target

Since the focus of this case is on "user retention", the RARAR model is adopted (based on the AARRR model and adjusted according to actual business needs).

Phase 1: User Retention

Goal: Provide value to users and keep them coming back.

Analysis: What are the factors that influence users to complete key behaviors? Whether the existing follow objects of the APP have met the user's needs; Are there other objects that users want to follow, but the "Follow" function is not open; What are the psychological factors that users turn on "Follow_Push".

Phase 2: User Activation

Purpose: Ensure new users see the value of your product when they first start.

Analysis: What are the psychological factors that influence user decision-making? How to make users aware of the value of key functions as early as possible? How to establish the relationship between product utility and users?

The third stage: Referral/Acquisition

Purpose: Let users share and discuss your product; encourage old users to bring in new users.

Analysis: How to trigger the user's sharing chain through "follow"; what are the psychological decision-making factors for the user level to migrate from "user" to "recommender"; how to find incremental space through key behaviors.

The fourth stage: Commercial realization of revenue

Purpose: To achieve commercial value.

Analysis: Improve retention through key behaviors and drive commercial monetization.

Key behavior - "Follow" process breakdown, among which 7 conversion rates are also influencing factors of the North Star indicator.

Step 4: Conduct the experiment

1. Content side

• Optimize "industry classification" and no longer use traditional classification methods to meet the needs of more users;
• More industries have been added, from the original 125 industries to 441 industries, covering a wider range of impacts;
• It is split into primary industry + secondary industry;

2. Functional side

• Come up with a series of user stories that allow users to have aha moments through the "Follow" feature.
• Users are divided into three categories: new users, users who were retained the next day after yesterday’s new users, and users who returned N days ago. Research is conducted on how to improve their retention rate through the product’s “follow” function.
• Shorten the "Follow" path to N steps, increase the number of "Industry_Follow" entries to M, open up "Push", strengthen guidance within the site, and use push/SMS outside the site to wake up users.
• After following an object (such as an industry, author, etc.), users can enable update push notifications on their own. Each step of adjustment requires restraint, and the impact will be gradually expanded. If the system automatically "pushes" messages after users "follow", users may feel that there are too many push notifications and feel disgusted, which is counterproductive.

Step 5: Analyze and summarize the experimental results

The users of the new version are divided into three categories: new users, users retained the next day from yesterday's new users, and users returning N days ago. After two weeks of observation and data analysis, both the North Star indicator and retention rate have been significantly improved.

How does user stratification work?

Users are stratified according to the North Star indicator. The formula of the North Star indicator "Follow_Daily Active" is broken down as follows:

"Focus" = "Industry_Focus" + "Author_Focus" + "Reporting Agency_Focus" + "Column_Focus" + "Question_Focus" "Daily Active" = "Daily New Users" + "Number of Users Retained the Next Day for Yesterday's New Users" + "∑Number of Returning Users in the Previous N Days (N>=1)"

A user feedback survey was conducted on the new version, and 64.33% of the respondents rated the push experience of the new version as "very good". In particular, there are only three reports that can be downloaded without any threshold every day on the APP. Through the report update push of the APP, they can download them in a timely and effective manner so that they can be used at any time during work.

It has been 5 months since this version was launched. After retrieving the retention data for analysis, the retention rate still remains at a good level.

Before the revision, the APP retention rate was slightly lower than that of APPs of the same scale in the same industry. Five months after the revision was launched, as shown in the figure below, the APP's next-week retention rate was about 20% higher than that of APPs of the same scale in the same industry.

Of course, there are many factors that can affect retention in these five months, not just this version. However, this version has achieved a retention rate of more than 10% in the first two weeks after its launch, and its retention rate has continued to increase to this day, far surpassing the same-sized apps in the same industry. This is enough to show that this version iteration is worthy of reference and reference, which is why I am sharing it this time.

Step 6: Self-reflection

1. The Pareto Principle

80% of iterations are trial and error/useless work, and only 20% are successful. Only by summarizing the experience of failure, reviewing the successful methodology, quickly iterating and trial and error, and finding the entry point/breakthrough point for growth can we find the best growth strategy.

2. Complete data system

The success rate of growth is positively correlated with the perfection rate of the data system. Comprehensive tracking, a complete and flexible reporting system, and a professional BI team are all very important.

3. Company support and investment

Growth hacking is not something that any employee can do on a whim; it requires the company boss to have a data-driven mindset. The company is willing to invest the cost to do this and has to be patient.

4. Choice of A/B Test

The cost of A/B TEST is greatly affected by the platform and technology maturity, so you need to choose the solution carefully at the beginning.

5. Multi-platform layout

APP web pages, H5, WAP, public accounts, mini programs, Toutiao accounts, Douyin accounts, etc., judge and consider the platform that is suitable for your own layout, form a multi-platform matrix, and use each platform to accumulate traffic.

6. Model construction and analysis

Our team still has room for improvement in using the RMF model and churn model to stratify users and conduct refined operations.

Author: DoraPM

Source: DoraPM Product Manager

Related reading:

How to build user churn warning?

4 ways to prevent user churn!