Content-based paid products: How to analyze the problem of “paying user” loss?

For operators of paid content products, the data changes of our "sponsors" are of great concern to us. When the churn rate of paying users shows an upward trend, how should we analyze the problem of churn of “sponsor daddies”?

Imagine your product is a city-state and your paying users are your citizens.

Every day new citizens join the city and some leave. In order for the city-state to grow and develop, in addition to paying attention to the number of new citizens, the reasons why citizens leave the city cannot be ignored.

Especially when the number of citizens leaving the city suddenly increases, it must be a warning that there is a problem in some aspects of your city-state. If it is not resolved, more citizens may leave. As the leader of a city-state, how should you face this crisis?

The operators of paid content products (such as iQiyi, QQ Music, etc.) and the data changes of their "sponsors" are of concern to us. When the churn rate of paying users shows an upward trend, how should we analyze the problem of churn of “sponsor daddies”? To help his "city-state" resolve the crisis.

This article follows the following ideas and uses actual cases to illustrate:

1. Clarify the issue of paid user churn;
2. According to the "Paying User Churn Hypothesis Tree", propose hypothetical reasons;
3. Based on the assumed reasons, propose adjustment and optimization plans and implement them;
4. Verify the hypothesis based on the effect and further analyze the cause if necessary.

【Case】

A membership-based (monthly, quarterly, annual membership) TV music product, whose main contents include audio, MV, live concerts, etc. The total number of users and paying users are growing steadily. Problem: The paying user churn rate suddenly increased by nearly 10% in October.

Then its analysis ideas are as follows:

1. Clarify the issue of paid user churn; what is the current definition of paid user churn rate? If we extend the time period to one year, is the current churn rate within the normal range? To what level does leadership want the current attrition rate to drop? etc.

2. Propose important hypotheses about causes; based on current data and user feedback, what are the biggest factors that affect the churn rate of paying users?

3. Develop optimization plans based on the assumptions. Based on the causal hypothesis, what are the factors we can control? And formulate corresponding optimization plans.

4. Verify the hypothesis based on the results. After the implementation of the plan, check the data again to verify whether the expected reduction in the churn rate of paying users has been achieved. Assuming not, we need to further analyze the reasons on this basis.

1. Clarify the issue of paid user churn

Clearly defining the problem can help us get closer to the real cause more effectively, especially for more complex operational issues such as paid user churn. Maybe we only discovered the problem through data, user feedback, etc. at the time, and we still need to further clarify the problem.

This can be clarified by asking the following 5 questions:

1. What is the background of the paid user churn problem? What stage of the life cycle is the product in? By whom was it proposed and based on what evidence (data analysis, user feedback, etc.)?
2. How are paying users and paying user churn rate defined? This step is especially important for students who are new to the project operation, as different types of products have different definitions of "churn".
3. To what level do the leader or yourself want this indicator to be optimized? In other words, once a certain standard is reached, the problem will be solved. Is your paying user conversion rate falling back into the normal range, or is it about to fall further below the average?
4. Are there any "hard flaws" such as bugs? Have there been any major bugs recently that affect paying users? Are the statistics accurate and are the standards consistent? This step is easy to overlook. If you don't do this check, the subsequent analysis may be in vain.
5. Determine the scope of the problem: If the overall user churn rate of a product is significant, or even exceeds the churn rate of paying users, then the problem may not only be limited to paying users.

【Case】

A membership-based (monthly, quarterly, annual membership) TV music product, whose main contents include audio, MV, live concerts, etc. The total number of users and paying users are growing steadily.

Problem: The paying user churn rate suddenly increased by nearly 10% in October.

This is the information we have currently, so how do we define this issue clearly?

1. Problem background: This problem was discovered in the weekly data report. The churn rate has been increasing rapidly in the past week.

2. Definition of concepts:

Paid users: Users who have purchased a membership (monthly, quarterly, or annual membership) and are within the validity period of the membership.

Paid user churn: users who have purchased a membership but no longer choose to pay more than 7 days after the membership expires.

Paid user churn rate = number of paid users churned / total number of paid users 7 days ago.

3. Standard for solving the problem: Leader requires that the paying user churn rate be reduced to a normal range of 5% to 10%.

4. Investigation of other factors: We have contacted product managers and R&D colleagues to confirm that there are no major bugs that affect key data recently and the data is accurate.

5. Scope of the problem: The overall user churn rate has changed little, but the paid user churn rate is significant.

2. Based on the “Paying User Churn Hypothesis Tree”, propose hypothetical reasons

The paid user churn hypothesis tree uses the "logic tree" to list the paid user churn problems in layers. This will help us to determine the approximate range of possible causes of loss based on the existing information.

Here are the three steps to find the cause of your hypothesis using a hypothesis tree:

1. Draw a hypothesis tree around the reasons for the loss of paying users

We can roughly divide the reasons into three parts: products, users, and external environment.

(1) Product changes

1. Functionality: Have the main functions changed recently? Especially features that can affect paying users. If so, it is necessary to evaluate the relevant data changes of the function and the impact on paying users.
2. Content: Whether there have been significant changes in the direction and quality of the content in the recent period. For example, on a certain video platform, a high-quality program brings a certain number of paying users to the platform. When the program is finished, if the quality of other content is too different, the paying users may be lost.
3. Experience: Whether factors such as UI changes and bugs affect the user experience.

(2) User changes

1. The user population or needs have changed. For example, the user source channel. Although a certain channel can promote the first payment, it may be because the users of this channel have a low match with the target users. There will also be a loss of paying users.
2. The user's experience when using the product. By sorting out the usage process of paying users and analyzing the data of each step or link, we can see which page or function has the loss problem.
3. There are new assessment options. Whether other alternative competing products emerge.

(3) Changes in the external environment

1. Key event nodes such as seasons and festivals;
2. Competing products compete for paying users;
3. Entry restrictions, such as WeChat restrictions, etc.

The above hypothesis tree is drawn only based on my personal experience. If your product has its own characteristics, you can also expand on this basis.

Of course, if you are very familiar with an industry or product, you can ignore this step and go directly to the next step.

1. Based on "data analysis + user return visit", we can identify the approximate cause range.

After drawing the hypothesis tree of the reasons for churn, you still need to do data analysis and user revisit, and use this as a basis for defining the scope of the hypothesized reasons.

What we need to pay attention to is that for the more complex issue of paying user churn, if we only do one of data analysis and user revisit, it may not be enough to get us closer to the real cause of churn.

1. Prioritize causal hypotheses

If there are multiple reasons, you also need to evaluate the priority and clarify which reason is the most important one? This will be used as a focus in subsequent optimization actions. Prioritization is especially important when time is urgent and energy is limited.

The above steps can provide a preliminary hypothesis of the cause; so why just propose a hypothesis first, instead of analyzing and verifying each influencing factor one by one?

The main reasons are:

1. Focus on the efficiency and quality of problem solving;
2. This problem itself is a complex one with multiple factors influencing it. According to the 2/8 principle, it is necessary to find the most critical influencing factor.

【Case】

A membership-based (monthly, quarterly, annual membership) TV music product, whose main contents include audio, MV, live concerts, etc. The total number of users and paying users are growing steadily. Problem: The paying user churn rate suddenly increased by nearly 10% in October.

The user return visit results and data analysis are as follows:

Since these are lost users, user interviews are difficult to conduct in practice and the content collected is limited. It is only used as a reference for cause analysis.

Conclusion: These may be paying users who have recently entered through new channels or activities, and their match with the target users is low.

Basis: After October 8, the loss of paying users was quite serious. These users placed orders at a similar time (around the beginning of September), and at this point in time, they were in the process of attracting new customers and expanding new channels, and the activity to attract new customers was to draw tickets for surrounding tours. It is possible that the users of a campaign or a channel are not the target users.

The reason assumes that the tree and range are as follows:

Prioritize these three reasons as follows:

1. Changes in the crowd: Last month’s new customer acquisition activities and channels may not have brought in real target users.

2. The content of the new customer acquisition activity uses the gimmick of "drawing a back-to-school season gift pack" which may result in fewer target users being obtained.

3. It may also have an impact before and after the end of summer vacation.

3. Develop an optimization plan

For possible causes, we need to quickly develop optimization plans to verify whether our cause hypothesis is correct.

During the formulation process, several aspects need to be paid attention to:

1. According to the priority of the hypothesized causes, decide which part needs to be focused on in developing an optimization plan.
2. Focus only on the parts of the hypothesized cause that you can control. For example, uncontrollable factors such as seasons and policy changes can be ignored for the time being and don't worry too much about them.
3. If it is difficult to formulate a plan directly, you can further break down the reasons and find the most important part of them. For example, experience factors can be divided into visual experience, usage fluency, text clarity, etc.

A membership-based (monthly, quarterly, annual membership) TV music product, whose main contents include audio, MV, live concerts, etc. The total number of users and paying users are growing steadily. Problem: The paying user churn rate suddenly increased by nearly 10% in October.

Assume the reasons are as follows:

1. Changes in the crowd: Last month’s new customer acquisition activities and channels did not bring in real target users.
2. The new customer acquisition activities use the gimmick of "drawing a back-to-school season gift pack" which may result in fewer target users being obtained.
3. There may also be an impact around the end of summer vacation.

Program Development Strategy

Focus on hypothesis 1; analyze recent new user source channels, evaluate the paid user churn of each channel, and abandon paid user source channels with high churn rates.

4. Verify the hypothesis based on the results

Due to the above steps, the optimization solutions we obtained are all based on the cause hypothesis. We still need to further implement the plan to evaluate whether the data effect has met our expectations.

If expectations are not met, you need to go back to the cause analysis and, if necessary, go back to the problem definition. Analyze the reasons for paying users again. Of course, under normal circumstances, 1 to 2 analyses are enough to reveal the reasons for the loss of paying users that are closer to the truth.

【Case】

A membership-based (monthly, quarterly, annual membership) TV music product, whose main contents include audio, MV, live concerts, etc. The total number of users and paying users are growing steadily, but the problem is that the paying user churn rate in October suddenly increased by nearly 10%.

After the optimization plan was implemented, it was found that the user churn rate gradually dropped to 6% in late October, basically achieving the expected goal.

Because different products have different specific forms and stages, and are limited by my current experience.

In actual operation, you may also need to pay attention to the following issues:

1. Differences in hypothesis trees: The hypothesis tree of reasons for loss of paying users can be continuously expanded and improved according to the characteristics of your own product. This will help you quickly determine the approximate scope of the cause in subsequent analysis.
2. The role of telephone follow-up: The actual effect of follow-up visits to paid users who have lost them remains to be discussed. Most users are unwilling to cooperate. Even if they are willing to cooperate, they may not obtain more real information in actual communication, even though they know the "5 why" analysis method.
3. "Hard flaws" cannot be ignored: When clarifying the problem, it is necessary to clarify whether the data is normal, whether there are important bugs affecting the data, and other "hard flaws". Only after these problems are eliminated will the subsequent analysis be meaningful. It is easy to overlook this step in actual operational work!

The above is a way of thinking about the problem of paid user loss, for reference by operations colleagues. In actual operations, you also need to build and improve your own thinking model based on your industry, product characteristics, job responsibilities, etc.; to help you improve your efficiency in analyzing and solving problems.

Source: Awei