20,000 words of practical information: How to build a user membership system from 0 to 1

There are many articles about membership systems, but most of them are specific methods and practical operations. Even some articles from 0 to 1 lack a global concept and a set of overall framework logic. The author of this article uses the membership system relationship diagram to give you a holistic perspective on the membership system, starting from the source of user behavior to the final level of rights and interests, giving you a complete thinking.

Li Dan once said: Everyone can do a 5-minute talk show. This sentence is also applicable to the membership system, and everyone can talk about something related to the membership system.

For example, how to classify users, user tags and CDP, how to obtain and consume membership points, membership levels and membership rights, and paid membership. But what exactly is a membership system? How should a membership system be constructed? Many people do not have a global concept and lack a set of overall framework logic.

Here I will introduce the above membership system relationship diagram, which will give you an overall perspective on the membership system. The arrows in the above diagram represent the direction of the relationship. For the 6 modules in the above diagram, user behavior is the source and the origin of subsequent modules 2-6. If the user does not have any behavior or action in the product system, then modules 2 to 6 will have no source.

Most of the current Internet C-end products basically have modules 12346, while module 5 is optional for some products. Some of the views described in this article are also partially applicable to B-end products and some offline formats, such as the catering and retail industry, and the education and culture industry.

For the sake of convenience, the terms "user" and "member" will be used interchangeably below. At the same time, the examples in this article will mainly focus on the commonly used APPs in Internet C-end products, please be aware of this.

1. User Behavior

First of all, before we do something, we should think about why we want to build a membership system.

If you don't know how to answer this question right now, let's think about it from another angle: If there is no membership system, what impact will it have on the product? What impact will it have on the company? Are these impacts good or bad? How wide and deep is the scope of the impact?

To illustrate my point, let me give you two examples.

Let's start with Kingsoft Office's WPS software. In the early years, WPS was a pure tool-based PC software that competed with Microsoft Office for more than 30 years. At that time, WPS did not have a membership system and could only earn revenue by selling software CDs to B-end customers, namely government departments and state-owned enterprises, and C-end customers contributed very little revenue.

After all, in the market environment at that time, with insufficient awareness of intellectual property rights and rampant piracy, most individual customers just searched for cracked and green versions online, and few users paid to buy them. When I joined Kingsoft Software through campus recruitment, I also gave each of our contracted students a gift box of WPS.

Later, WPS began to try to transform into the Internet, started the PC + mobile + cloud service model, and began to make money for WPS members through value-added services. This business model is running very well, and by 2023, the annual revenue will reach 4.556 billion yuan. If WPS is still a stand-alone version now, can it achieve this performance?

In fact, when it comes to Kingsoft Software's transformation to mobile Internet, Kingsoft PowerWord was attempted earlier than Kingsoft WPS. Around 2008, Kingsoft PowerWord's general manager Li Wanqiang (yes, the same Ali who later became Xiaomi) proposed user thinking and Internet thinking when he was working on Google-Kingsoft PowerWord.

In simple terms, the software used to generate revenue by selling software licenses through offline channels, but now the basic functions of the software can be downloaded for free, and then the value-added services are charged. For example, the Google-Kingsoft PowerWord software is free, and some dictionaries/vocabularies are pre-installed in it. Users can purchase various authoritative and professional dictionaries online as needed.

This concept seems ordinary now, but in 2008, 99% of the software on the market made money by selling licenses. Kingsoft PowerWord's innovative business model was ahead of its time. Later, Kingsoft WPS also tried this model and it went very well.

The key point of this model is the creation of a membership system. That is, you must have an account system to record various user behavior data and consumption purchase data, with a series of functions such as tags, levels, points and rights as supporting functions.

Let's take the case of Meitu XiuXiu. In the early years, people mainly used Photoshop (abbreviated as PS for convenience) when editing photos. However, those who have used this software know that the threshold of PS is too high and the operation is too difficult. In the early years, there were many PS video tutorials on the Internet, and there were also many PS books with CDs and source files in bookstores. Later, Meitu XiuXiu and Photoscape came out, and quickly accumulated a large number of users with their ultra-low threshold and simple operation.

You may have guessed that in the early years, both Meitu XiuXiu and Photo Magic were pure PC client tool software like WPS, without an account system or membership system, so they were powerless even if they wanted to commercialize. Later, after having a user system, they began to slowly try to commercialize, becoming paid members, making money from advertising services and value-added services. According to Meitu's financial report, its annual revenue in 2022 was 2.085 billion yuan.

Many years ago, someone asked me a question: I can understand why JD.com and Taobao have account systems, but why do all kinds of tool software, such as antivirus software and cleaning assistants, start to have account systems? The above explanation should be able to answer this question, because only with an account system can user behavior be recorded, which is Module 1 in the relationship diagram at the beginning.

Without an account system, there will be no module 1. When a user changes computers or mobile devices, all his previous behavior records will be lost, and no subsequent user and business value can be generated. This also explains why many products always guide you to log in or register when new users enter. At the same time, in order to lower the operation threshold for new users, there are verification code login, WeChat login, one-click login, etc.

Of course, some readers may ask, if there is no account system, I can make money through advertising, such as websites like 17173, Duowan and Sina. There is nothing wrong with this, but if you have an account system and a membership system, do you have more ways and paths for commercial monetization?

Taking Bilibili as an example, its 2023 financial report shows that its advertising revenue was 1.64 billion yuan, accounting for only 21%. More revenue came from value-added services and gaming business. The key point of these two businesses is to have a membership system.

Therefore, the account system/membership system that we are accustomed to now, as well as the membership levels/membership points that we are familiar with, did not exist at the beginning. These are the products of product evolution and commercialization.

Back to the original question, the reasons or business value of building a membership system can be briefly summarized as follows:

• Extend user life cycle.
• Maintain a long-term and stable relationship between products and users.
• Allow users to generate as much commercial value as possible.
• Improve user activity and loyalty through differentiated services and marketing strategies.
• Improve efficiency and increase revenue for enterprises (products).

I will elaborate on the above business value in the following content, so I won’t expand on it here. You can simply understand it as follows: the essence of the membership system is to invest limited resources in user groups that are expected to generate more revenue.

Let's go back to the topic of this section: user behavior. As the name suggests, user behavior refers to the behavior of users on the product. For example, I open the Tencent Video APP, search for the name of a TV series or movie, or go to my page to view cached data, and then set the height of the bullet screen, set the multiple playback, etc. These are my user behaviors.

After recording user behavior, user data can be generated, such as the proportion of customers adding items to the shopping cart through various paths such as [home page recommendations, category search, search, frequent purchases, favorites and historical orders] in the e-commerce app, and the frequency and time of users using each functional module of the product.

Then, based on these data, we can make subsequent membership system planning, such as user levels, user points, user tags and paid memberships, so as to better realize the commercial value of the product and the company.

The content about user behavior data analysis will not be expanded here. If there is an opportunity later, I will write a separate article to discuss this content.

2. User Level

Still the same, why do we need to build user levels?

How do you manage a country with hundreds of millions of people? Our ancestor Qin Shihuang gave us the answer: the county system. Now it has evolved into a five-tier administrative governance structure: province-city-district-township (street)-village (community).

The same is true for products. When your product has a large number of users, such as 30 million, how do you manage them? How do you operate users? If you do not differentiate between them and treat them indiscriminately, how can you operate and manage these 30 million users? How can you use the best steel on the blade?

Therefore, in order to maximize commercial value, we need to manage users in a refined manner. In order to achieve refined management, we need to classify and grade them in the same way as the administrative governance structure.

1. User stratification

I believe everyone has heard of the Pareto Principle. So, can we divide these users into two camps or groups, and then invest 80% of our energy and resources in 20% of our customers, so that they can generate 80% of the revenue? It is feasible in theory, but it does not work in practice. Such a rough method seems too hasty and perfunctory, and it needs to be segmented.

When you are planning and operating a product, you need to first consider how the product can be divided into different levels and grades. This is the first and most important issue that product managers and operators need to consider, and it is also one of the necessary skills. It is difficult to understand all users at once, but it will be much easier to divide and conquer. The direct effect of dividing user groups is to better understand the target users of the product and the market competition.

• What types of user groups are there?
• What are the characteristics of these user groups? How many user groups are there?
• What user groups do competitors cover?
• Which market segments are unfilled? Which market segments are the must-win areas for long-term goals?
• What is the competitive strategy for a product from inception to maturity? Which user group should be targeted first and which one next?

User stratification is a very important step in the membership system, module 3, or user level. Without user stratification and user level, subsequent modules 3456 will not be established. But how to stratify users? Different product forms and industry standards are not consistent. E-commerce transaction products can be divided by RFM or transaction amount, but reading information products such as WeChat Reading may be more suitable for reading days and reading time.

Maoyan Movie divides users into the following levels. Each different user layer or user group plays a different role, and there are also connections and interactions between groups. For example, contributing users and active users, ordinary users and professional users, users in different fields, pretentious users and supporting users. These user groups are not long-term and vary according to the stage of the product.

NetEase Cloud Music divides users into the following groups. Its grouping factors can be divided into two categories: basic demographic attributes (such as age, gender, education level, occupation, etc.) and vertical field attributes (in the case of music, it is the degree of music preference). To divide user groups using a two-dimensional coordinate system, two factors are sufficient.

The above-mentioned user stratification method of NetEase Cloud Music is quite useful in product planning and operation and promotion, but this stratification method has a major disadvantage or drawback: it is not quantitative and not intuitive enough. After we stratify the users of NetEase Cloud Music according to the above stratification method, there may still be hundreds of millions of users who fall into the mass user level, which still does not solve our problem of refined operation.

Therefore, stratification alone is not enough, and grading is also required.

2. User classification

The C-end products currently on the market can be roughly divided into two categories. One is that users will make transactions in the product, such as JD.com, Taobao, and Pinduoduo, that is, users pay money. The other is that users spend energy in the product, such as Toutiao, WeChat, Douyin, and Xiaohongshu, that is, users spend time. Of course, there are also some that have both, such as I have a membership to listen to songs on NetEase Cloud Music and I have a membership to watch TV series on Tencent Video. We will not go into details here.

For the first category, we can use the amount of consumption as the basis for grading, and for the second category, we can use the duration of use (frequency * single duration). In the membership level system, we should consider making the grading standards as simple as possible to reduce the user's conversion cost as much as possible, so that users can upgrade more clearly.

Haidilao belongs to the first category. The main basis for its grading is the amount of money spent by customers in the store, and activities (activity) are added to the growth value calculation system of the membership level. It is worth noting that while customers accumulate growth points when spending in Haidilao stores, they can also accumulate Laobi coins (which can be regarded as points). This practice is similar to the consumption return of Jingdou in JD.com.

JD's membership level is calculated based on the Jingxiang value, which is calculated by account points + consumption points + active points. Account points are mainly used to guide users to authenticate their real names/bind their email addresses/bind WeChat accounts/bind their QQ accounts. Consumption points are not expanded as the name implies. Active points are mainly used to guide users to log in/evaluate/post their orders/answer questions, etc. However, because the brilliance of JD Plus is too dazzling and the rights and interests of JD's level members are too tasteless, the presence of JD's level members has always been relatively low.

The following is how NetEase Cloud Music determines the user level. The only criteria for grading are the number of songs listened to and the number of days logged in. The tricky thing here is that NetEase Cloud Music uses the number of songs listened to instead of the length of time listened to. The operator may also guess that there are many users who repeat a single song or a list, and NetEase Cloud Music uses the number of songs listened to instead of the length of time listened to in the grading in order to allow users to discover more music.

To sum up : For enterprises or products such as fast-moving consumer goods retail, catering, etc. that mainly engage in product and service transactions, the following dimensions can be considered comprehensively when dividing membership levels and divided according to growth values.

• Affordability: Purchase frequency, purchase quantity, purchase category, purchase amount
• Activity: login, evaluation, order sharing, forwarding, referral, product recommendations, etc.
• Member quality: real-name authentication, returns and exchanges, financial credit, etc.

User stratification and classification complement each other. Some products skip classification and do classification directly, while others confuse the two. Classification is relatively clear and easy to understand, but the definition of stratification is a bit vague and the standards are not unique. For example, in the figure below, the new product preference customers and high net worth standards, the product department believes that high net worth is a customer with purchase frequency > X and consumption amount > Y. The operation department may think that it is necessary to add limiting conditions, such as after registration or in the last XX days.

I am not sure whether the above-mentioned grading system in NetEase Cloud Music is for all users or just for the general public. I am not sure whether there is a set of user grading standards and algorithms for opinion leaders or industry practitioners within NetEase Cloud Music. If I were an operator, and if the number of opinion leaders exceeded 200, I would think it would be necessary to grade them.

After you have determined the user classification standard, you still have to consider how many level segments you want to set. For example, NetEase Cloud Music has 500 million users. You have decided to use the number of songs listened to + the number of days logged in as the level classification standard, but how many segments do you plan to divide these 500 million users into? 10 or 30? You should know that the highest QQ level is already 202.

The conventional membership system should have about 7 levels (that is, 3 levels up and down, and 4-10 levels in the end. Haidilao is 4 levels, JD.com is 5 levels, Bilibili is 6 levels, and NetEase Cloud Music is 10 levels). In the field of psychology, 7 is an incredible number that scholars call. Most people's short-term memory capacity is only 7 at most. If it exceeds 7, they will forget. Therefore, most people classify the memory content into seven units. The same applies to the issue of user classification. Too many levels are not convenient for users to understand, and too few levels are not conducive to focusing on top users.

Then the question arises again, are these 7 levels or levels evenly distributed, normally distributed, or distributed in a pyramid structure? Still applying the Pareto Principle, I personally think that the pyramid structure of the level distribution is the most appropriate.

Based on the number of levels determined above, we can then segment all users into small segments and calculate the number and proportion of users in each segment. For example, if we take 100 yuan as a step, we can calculate the number and proportion of users in each amount segment of 1-100 yuan, 101-200 yuan, 201-300 yuan...

Then, according to the 80/20 rule, the amount segments are merged. In this way, the dozens or hundreds of amount segments we split above will be merged into the final 7 or so, and the final level interval segments can also be finalized. The following figure shows the level threshold value division and the estimated number of people in a certain product. The grading of this product basically conforms to the pyramid structure.

Many product and operation personnel directly skip the above analysis, deduction and calculation processes, or directly refer to and apply the grades and classification standards of competitors. However, if this is a new product (although the probability is very low) and there are no competitors to refer to, you can use the above method to stratify and grade.

3. Relegation and relegation

We have described above how to improve the growth value and thus improve the user level. There is a key modifier, that is, the consumption records of the past XXX days are used to calculate the growth value. Haidilao's XXX is the past 12 months. For example, on May 1, 2023, the consumption amount of the previous 365 days will be pulled and the growth value will be calculated to determine the membership level, because the department team building used my membership account to order 6,000 yuan to meet the standards of Jinhai members.

Then I can always enjoy the membership benefits of Jinhai members? No, most user membership levels will be calculated and updated regularly, some may be updated daily, some weekly, some monthly, some quarterly or annually, but the level will be updated. The specific update cycle needs to be analyzed based on the specific situation of the product and the company.

If my growth value is less than 6,000 points when I recalculate, I will drop to Silver Sea Member. So in order to continue to enjoy the rights and benefits of Gold Sea Member, I need to frequently patronize Haidilao or participate in activities to gain growth points to maintain my membership level.

From the perspective of users or consumers, the retention and downgrade system is unreasonable, but from the perspective of products and operations, it is a last resort. If the growth value of the most recent XXX days is not taken, and the modifier of XXX days is not added, the pyramid structure of the entire membership level will change, which goes against the original intention of designing the user membership level.

So this also gave birth to a new business : lending (selling) Black Sea memberships on Xianyu.

When downgrading, some members can be directly downgraded to the next level, such as Gold Sea Members becoming Silver Sea Members. Some members can be directly downgraded to multiple levels, such as Black Sea Members directly downgraded to Red Sea Members. Similarly, which method to use should also be analyzed and demonstrated based on the actual situation of the product and the company.

3. User Points

Before starting this section, there is still a question: Why do products need points, or what is the significance of points?

Many companies or products, without thinking through the above question, have added points to their product system (or membership system), and explained the above question with the logic of "increasing user activity and loyalty, and increasing user participation and sense of achievement." But does this logic work?

• Improve activity: I am active on Zhihu because I can learn a lot of interesting things, because I have questions that need to be answered, and because I want to build my personal brand and influence on Zhihu, not because of the points. Baidu Tieba gives me double the points, but will I go there to solve my problems?
• Improve sense of accomplishment: I am a heavy user of Bilibili and NetEase Cloud Music. I recognize and agree with the sense of identity and accomplishment that user levels such as LV6 bring me, but this is the result of a comprehensive evaluation of user growth value and activity value, and is not entirely brought by the so-called points.

Back to the original question: points can improve user loyalty and activity of the product to a certain extent, so points may be needed.

However, whether a product requires points needs to be analyzed specifically based on the product’s form and user scale.

For example, for information reading or music products, I personally think that it is more important to improve user levels than points, while e-commerce trading products are more suitable for introducing points as a virtual currency. This is because points can make customers feel that they have extra benefits and have a sense of getting a bargain. What users need is not necessarily getting a bargain, but the feeling of getting a bargain.

For example, if the price and delivery time of a certain hot pot base are the same on JD.com and Haidilao, if Haidilao gives Lao coins (points) but JD.com does not, I think rational customers may be more inclined to buy it on Haidilao in order to get Lao coins (Lao coins can be used as cash or exchanged for goods). After all, it makes users feel that they are getting a good deal.

However, not all transaction products will have points. For example, Meituan does not have points. The reason is very simple: there is no input-output ratio. Based on Meituan's annual transaction volume of more than 1 trillion yuan, if 1 point is accumulated for every 100 yuan spent, and 100 points are equivalent to 1 yuan, theoretically it means a point cost of 100 million yuan. Although this cost only accounts for 0.26% of Meituan's 38.7 billion yuan operating profit in 2023, given Meituan's market share, there is no need to use points as a means to increase user loyalty.

Some readers may say that JD.com has a points product, namely Jingdou, and JD.com's annual GMV has exceeded one trillion yuan. But if you look closely at the relevant introduction of Jingdou, you will find that the return rate of Jingdou starts at 5%, the exchange rate between JD.com and RMB is 100:1, and there are many situations where Jingdou is not returned. Calculated based on the 8 trillion GMV, the overall theoretical cost limit of Jingdou is only 4 billion yuan. Compared with JD.com's annual revenue of 1084.7 billion yuan, the theoretical marketing cost of Jingdou accounts for 0.37%, and the overall cost is relatively controllable.

I checked Haidilao's 2023 financial report. Its stores had annual revenue of 39.2 billion yuan and a net profit of 5.8 billion yuan. The marketing cost for Laobi coins, or points, was 70.01 million yuan, and the points cost accounted for 0.2% of the annual revenue.

The entry threshold of points is not high, but the threshold after entering is very, very high. Points are debts, and not everyone can afford to play with them. Moreover, the most difficult thing about points is that the input-output ratio is not easy to calculate, and the balance between points output and consumption is not easy to control.

If the above account can be calculated clearly at the product or operation level, and the input-output ratio is in line with expectations, then the points function can be implemented. If it is not clear, it is not recommended to start it hastily.

Here we explain the difference between growth value and points so that everyone can better understand the concept of user points.

• The growth value can be regarded as the measurement value of all business actions of the user and is the criterion for determining the user level. User points are points.
• Growth points can be obtained through consumption or other user behaviors. The way to obtain points is similar to growth points.
• Growth points are not consumable and will only be automatically consumed when the user's level is maintained or downgraded. There is no other consumption outlet. Points are consumable and can be used for consumption deductions, participation in activities and product redemption.
• Points are debts, that is, all points issued by the product must be exchanged for users at some point in the future, and this exchange process will involve costs. The growth value is just a measurement value, although it may be partially or completely cleared at some point in the future, but there is no need to pay additional costs.
• The entry threshold for points is not high, but the threshold after entry is very, very high. The starting threshold for growth value is relatively high, but the threshold after entry is relatively low.
• The most difficult part of points is that the input-output ratio is difficult to measure, that is, the balance between point output and consumption is not easy to control. The most difficult part of growth value is how to set the level rights and interests corresponding to the user level.

Assuming that we have confirmed that we want to add the member (user) points function to the membership system, we must consider the path of point acquisition and point consumption. Point issuance determines the scale of the points system and the growth rate of the stock point cost, while point consumption determines the influence of the points system on users.

As we said before, the essence of points is a debt owed by the company to users. In financial calculations, it is a provisioned cost, which means that the higher the cash cost corresponding to the existing points, the more disadvantageous it is for the company, because it means that the company needs to repay a high debt to consumers. Therefore, before designing the point acquisition method, we need to calculate the cost of points, then estimate the amount of points issued, and then split it into each different acquisition method.

Let's take the example of Haidilao, which we are familiar with. In 2023, Haidilao's store revenue was 39.2 billion and its net profit was 5.8 billion. Combined with the financial report data of Haidilao's boss in 2023, the cost of Haidilao's point marketing was 70.01 million yuan. According to the ratio of Laobi to RMB 20:1, in theory, it distributed 0.7*20=1.4 billion Laobi.

Combined with Haidilao's points (also known as Laobi) rules, one Laobi can be accumulated for every 1 yuan spent. According to the store revenue of 39.2 billion yuan, the theoretical number of Laobi issued is 39.2 billion Laobi. At this point, you should be able to find the problem. Theoretically, 39.2 billion Laobi were issued, but the actual Laobi exchanged was 1.4 billion, with a difference of 37.8 billion Laobi. Where did they go ?

A careful look at how Lao Coins are obtained reveals that customers must log in and indicate their membership status when dining at the store in order to obtain Lao Coins. The aforementioned 39.2 billion Lao Coins is the theoretical maximum that will be generated when 100% of customers log in to place their orders. In reality, the ratio may be 50% or something else. The exact amount may only be known by Haidilao itself.

Secondly, after customers, that is, members (users), obtain Laobi, not 100% of users will redeem 100% of Laobi, and I have no way of knowing what the ratio is. Youzan's solution consultant once gave a data: the point production and sales ratio or point redemption rate is in the range of 30%-50%, which is a relatively healthy and reasonable range. If it is too low, it means that the point perception is too low and the user's enthusiasm for participation is not high. If it is too high, it means that it is too easy to obtain points and the budget is wasted.

We can conclude from Haidilao's financial report data that the final ratio of Lao Coin generation to consumption is 3.6%, that is, the product of the ordering login ratio * points redemption ratio shown in the figure below is 3.6%. Assuming that 12% of users will log in to order food, it means that only 30% of the Lao Coins generated in this part have been redeemed or used, and 70% of the Lao Coins have expired.

From this, we can see where the difference of the 37.8 billion Laobi lies. The above 30%-50% points redemption rate is just an ideal value. Among the several projects I have experienced, the lowest redemption rate is only 1%, the highest is no more than 18%, and around 10% is normal.

Because the distribution of stock points is not even, for example, 20% of members may have more than 80% of the points, and among these 20%, can the proportion of actually using points to redeem points for various reasons be 50%? But I think the redemption rate of JD.com Jingdou should be much higher than that of Haidilao, after all, it can be easily selected on the order settlement page.

After inferring the number of Laobi that can be issued theoretically in a year according to the above method, we can then consider the channels and methods through which Laobi can be issued. For transaction-type products, the main channel of issuance should be transaction gifts, and the proportion should not be less than 50%, and not more than 80%. The following is the standard for issuing points and the cost calculation table in my previous project, for reference only (the calculation is based on 1 million members).

In the process of obtaining and consuming points, it is necessary to monitor the process and warn of behaviors. Relevant thresholds can be set, such as the upper limit of points that a single customer can obtain per day/week/single channel (consumption behavior can be set without setting points, and points can be issued at a fixed percentage of the user's actual payment amount) to prevent the behavior of hoarding points. If the threshold is triggered, manual intervention and identification can be carried out.

For the consumption of points, it is also necessary to monitor and warn of large-scale consumption. In addition to monitoring the large-scale consumption of points by individual members, it is also necessary to promptly detect whether there are system loopholes, points mall settings errors, etc. It is also necessary to monitor the use of points in all points consumption channels, detect abnormal problems as soon as possible, and intervene and intervene as soon as possible.

There is still a lot of room for manipulation when redeeming or consuming member points, and this space is the so-called premium. Simply put, the prices of goods that can be redeemed in the points mall are generally more expensive than those on third-party e-commerce platforms such as JD.com and Taobao.

Let’s take Haidilao as an example : for example, a power bank is sold for 100 yuan on JD.com, but the total cost price at Haidilao is 80 yuan (the power bank is used for convenience of calculation, and a kettle is used in the figure below), and the price in Haidilao’s points mall should correspond to 2,000 Lao coins, but the actual situation is that the power bank is sold for 2,300 Lao coins (corresponding to a value of 2,300/20=115 yuan). The extra 15% is the points (Lao coins) premium.

The original value of 1 Laobi was 1÷20=0.05 yuan, but now the value of Laobi is 100÷2300=0.043 yuan. The intuitive feeling is that Laobi is worthless. If calculated based on the total budget of 70.01 million, the corresponding number of Laobi that can be exchanged is 70010000÷0.043=1.628 billion Laobi, which is 228 million more than the expected 1.4 billion Laobi.

Some people may say that Haidilao is not authentic. If you sell a power bank for 100 yuan on e-commerce platforms such as JD.com and Taobao, you will sell it to me for 2,300 Laobi, which is 115 yuan. Most points malls do this, especially those credit card points malls, where the premium is even more outrageous. But users will feel that this Laobi is given to me as an extra gift from the product. Instead of leaving it here and waiting for it to expire and be discarded, it is better to pick up a few things that can be used.

At this point, some users may ask what the relationship is between the user level and the user points in the relationship diagram at the beginning of the article? In fact, it is very simple. When planning the user level of the membership system, some products set up a setting that user points will be given after the user level is upgraded. Strictly speaking, the given user points belong to the category of "user rights and interests", but it is a bit inappropriate to classify such a huge content as user points into user rights and interests, so I added this connecting line in the relationship diagram.

Finally, the first-in-first-out principle is generally followed when consuming points. It is not difficult in theory, but there are still some difficulties in product design. For example, I earned 100 points on May 1 and 300 points on May 3 respectively. On May 5, I consumed 200 points. Later, a partial refund was generated, and 180 points of the 200 points were returned. How to deal with this? Interested readers can figure it out by themselves.

4. User Tags

Before starting this section, let’s ask a soul-searching question: Why does a product need user tags, or what is the meaning/value of user tags?

• In some business scenarios, membership levels cannot meet business needs, such as new user activation, old user benefits, and silent customer awakening.
• The logic and dimensions of user level classification have some limitations and cannot meet the needs of precision marketing and refined operations. For example, products need to launch targeted activities for female customers, married customers, car owners, doctors or teachers, Android phone users, and customers in Jiangsu, Zhejiang and Shanghai.
• Customers can be divided into about 10 levels at most, which is not fine enough to carry out refined operations, improve conversion rates and ROI (return on investment), and will result in unnecessary waste of marketing expenses.
• Personalized service: In the field of customer service, user tags can help companies provide personalized service experiences for different users, such as customized recommendations, discounts, and care. Personalized services can improve user satisfaction and loyalty, and promote user repurchase and word-of-mouth communication.
• Market segmentation: User tags help companies segment the market into smaller, more specific target markets, thereby targeting potential customers more accurately.
• Data analysis: User tags are an important basis for data analysis. By mining and analyzing user tag data, companies can discover market trends, predict user behavior, and evaluate marketing effectiveness. This helps companies make more informed decisions and improve business operation efficiency and market competitiveness.
• Risk management and compliance: In some industries, such as finance and healthcare, user tags can help companies identify potential high-risk users or violations, so that they can take corresponding risk management measures, such as red lists and gray lists in addition to white lists and black lists.

In short, user tags are one of the important means for modern enterprises to achieve precision marketing, product optimization, personalized services, market segmentation, data analysis, enhance user engagement and risk management.

User tags can take many forms, such as natural attributes of users, statistical indicators of user transactions and asset data, or indicators summarized based on certain rules. Regardless of the form, they describe and characterize a certain dimension of the user.

The classification method of the above tags is too general and is inconvenient for overall planning and management. Each product can be classified in multiple levels according to its actual situation and divided into first, second, third and fourth levels. It can be further subdivided under each category. For example, the population attributes of the first-level classification can be subdivided into basic information and location information, and the basic information can be subdivided into age and gender.

A good label framework or classification and hierarchy method must be based on business scenarios. It should fully cover the user's behavior cycle and business workflow. For example, for e-commerce transaction products, we can first restore the business process, and try to cover the entire life cycle of the user or product, and then sort out the business scenarios that will be applied to the tags based on the business goals, and then set up a label system that matches the business.

We all know that the cost of obtaining new customers for e-commerce products is more than N times that of maintaining old customers, and this N is likely to be no less than 5. If we want to maintain old customers and do activation or recall of old users, then there should be corresponding old customer tags, sleepy tags and churn tags. Through the business application scenarios, we can infer what kind of user tags we need.

Next is the production process of labels. From the perspective of the operational level, there can also be different classification methods. The most common ones are: fact labels, rule statistics labels, and algorithm prediction labels.

• Fact tag: This is easier to understand. It is the basic information items for the user to fill in when registering or placing an order, such as gender, age, name, education, mobile phone number, residence address and work address, etc.
• Rule statistics tags: tags obtained through simple processing operations of statistical functions, such as: the most common Internet access time period, purchase cycle or frequency, average consumer unit price, brand preference, consumption power level, payment amount, order terminal preference (mini program or APP), payment method preference, etc.
• Algorithm prediction tags: use decision tree algorithms, regression algorithms, etc. to explore the relevant characteristics of users to label users, and then cooperate with different marketing strategies. Common ones include new product preferences, price-sensitive types, high net worth people, workplace elites, default probability, product loyalty, etc.

Below I will use a specific scenario example to help you deepen your understanding of label production and creation. Here I select the above rules to count the consumption power level tags in the label, and then briefly talk about the process of label production and creation.

First of all, the consumption power level is a label. When we plan and design this label, we must think clearly whether it is made into a qualitative enumeration value label or a quantitative numerical label? If it is the former, the enumeration value may be extremely high, high, medium, general and low. Then we need to divide the intervals of these 5 enumeration values. Is the extremely high threshold 10,000 yuan or 30,000 yuan? Is the extremely low threshold 100 yuan or 500 yuan? Is the statistical caliber to take the cumulative consumption or the consumption amount in the past XXX days?

If it is a quantitative numerical tag, then the consumption level tag will be somewhat duplicated with the scope of payment amount. When designing and defining the tag system, it is necessary to clarify the meaning and purpose of each tag, as well as relevant specifications and constraints, otherwise it will lead to ambiguity and confusion of the tags, and operators will not be able to tell which one should be used when using the tags.

Therefore, the consumption power level tag may be more suitable to use the qualitative enumeration value tag. The consumption power level is suitable for quickly screening customer groups to improve the awareness and operational efficiency of business personnel, such as inviting customers with [consumption power level = extremely high] to participate in offline product launch conferences. The payment amount value tag is suitable for more accurate selection of users through the tag, such as screening customers with consumption amounts of 500-1,000 to invite them to participate in the new product questionnaire.

Secondly, we need to consider where the data comes from, how it comes from, and how to process it after it comes. For example, the payment amount mentioned above, such as Haidilao sells hot pot bases in mini programs, APPs and third-party platforms. So when collecting the numbers, should we collect the full-caliber order data? Is it offline or in real time when collecting. The collected data needs to be cleaned, some refunds and after-sales orders are eliminated, and then structured processing is carried out.

When collecting omni-channel data, there is another difficulty. How to identify that Zhang San from JD.com and Mr. Zhang from self-operated e-commerce platform are the same person, or orders placed by the agency and orders placed by customers need to be classified into the same user. This will involve ID-Mapping technology .

In addition to being divided into fact labels, rule statistical labels, and algorithm prediction labels according to the production process, user labels can also be divided into static labels and dynamic labels according to the extracted dimensions.

• Static tags: determined during the data collection and collation stage, usually classified and marked based on fixed attributes or characteristics. These tags do not change with time or event, such as the user's age, gender, education, etc.
• Dynamic tags: used to capture and reflect dynamic changes in data, providing more accurate and real-time information. For example, the user's purchase frequency, the last purchase time, etc. are dynamic tags.

We have planned the user tag system based on various dimensions and standards, and then we can manage/view these tag systems through a visual page. For example, we can view the consumption power level mentioned above, the number and proportion of each enumeration value, such as the extremely high number of 2,500 people account for 4%.

The user tag system is constantly developing and iterating, and a new concept has been introduced later: CDP. This term was first proposed by marketing technology expert David Raab in 2013 (Customer Data Platform). Simply put, CDP is a consumer-centric data management platform, mainly aimed at marketing and operation personnel, and is used in scenarios such as refined and automated operations and advertising marketing.

The following is the CDP platform planning plan I made in my previous project, for understanding only.

For the user tag system, using tags to filter target customer groups and output crowd lists is the easiest and direct application method. The more common application scenarios include: thousands of people, product pricing, event customer circles and precise marketing.

Thousands of People : E-commerce products implement a personalized recommendation strategy with thousands of People through optimizing the label system, matching users' interests and preferences with products, and providing users with personalized recommendation results. This personalized recommendation strategy greatly improves users' shopping experience and satisfaction, and improves purchasing efficiency, conversion rate and order value.

Product pricing : In economics, pricing methods are divided into: first-level price discrimination, second-level price discrimination, and third-level price discrimination. First-level price discrimination is also called price per person, and Zhang San and Li Si purchase the same product with different prices. Second-level price discrimination is also called price per quantity, that is, the price of buying 10 pieces and buying 2 pieces is different. Third-level price discrimination is price per category, such as the price of working people and students purchasing MacBooks is different. User tags can provide user-level tool support for pricing products with first-level and third-level price discrimination.

Event circle customers : For example, e-commerce platforms have to do an event and send a batch of coupons recently, because the total denomination of the coupon is limited, it is impossible to achieve equal sharing of rain and dew. After the operation, the decision is to issue the coupons to some high-net-worth individuals. So the question is, high-net-worth individuals are a broad concept. Who is it? This can be combined through three user tags, and customers who meet the following three conditions will be selected.

If the number of people in this target group is too small, we can try to modify the value range of a certain tag, such as adjusting the number of visits to 30, or adjusting the sensitivity of the coupon to [Medium and High], or adjusting the consumption capacity to [High, High]. The same applies to various marketing activities such as full discounts and full bonuses.

In real applications, the above method is often used to form a customer group package by combining multiple user tags, and then directly select the corresponding customer group package during the event. For example, the customer group package of West China October New Customer is composed of three user tags: [Register Time = October] + [Register Channel = West China Fresh Go] + [Purchase Amount = 0 Yuan].

Precision Marketing : Many people have heard the saying "I know half of the advertising costs are wasted, but I don't know which half." This situation was common in the early years. With the development and progress of precision marketing, the proportion of this waste is gradually shrinking. In the past, everyone saw the same advertisement, but now they can set different Landing Page pages for different customer groups, set different ways of reaching, and set different registration rewards.

The above content is a label system for the user (member) level. In fact, it can also be established for stores or other roles, and then these label systems are used in corresponding business scenarios. The following is a partial screenshot of Mengniu's CDP system for reference.

5. Paid Membership

According to convention, there is still an opening soul-searching question: Why do you need to be a paid member, or what is the business value of a paid member?

Speaking of paid members in the domestic Internet industry, we must not ignore QQ members. The QQ red, yellow, blue and green diamonds in the early years played a great role in promoting the user education and market enlightenment of paid members, so when various products are launched in the later stage, users will be easily accepted.

Regarding the business value and necessity of paid members, I feel that everyone can say a few things, so I will also briefly analyze and summarize them.

• Industry trends : In recent years, C-end products have moved from incremental markets to stock markets. In order to stabilize the moat, enterprises and products will inevitably pay more attention to stock users, especially high-value users, otherwise users will easily flow to competitors.
• Source of income : The paid membership model can not only provide stable membership fee income, but also further increase the source of income by providing exclusive content and advanced services. For example, after a user purchases a VIP membership to get listen to books, a large number of users will also purchase an additional single course, and get listen to books and get the additional income from the course sales.
• Increase sunk costs : Because users spend money to purchase paid members, the sunk costs are increased. Users hate losses and think that if VIP members buy them, wouldn’t it be too wasteful? So the cost of users leaving will increase, the churn rate will decrease, and the retention rate will be relatively higher.
• Improve customer stickiness : By providing personalized services and discounts, the paid membership model can improve the loyalty and stickiness of members, making members more willing to stay on the platform for a long time and actively participate in various activities of the platform.
• Maximizing user value : Paid members and level members have made explicit distinctions, allowing enterprises to better identify and provide targeted better value-added services for paid members, thereby maximizing user value and making the growth of enterprises more efficient.

Paid members can be regarded as screening funnels for products or platforms, screening out high-value, high-sticking, and high-loyalty users, and then conducting refined operations to achieve more profitable purposes.

• 3% of Alibaba and Taobao’s top members contributed 70% of their revenue flow.
• According to data provided by Alibaba, the purchase orders purchased by 88VIP members are twice that of ordinary users, and the purchase width is 6 times that of ordinary users.
• According to data released by JD.com, 65% of PLUS members are young people under the age of 35, and nearly 60% of members live in first- and second-tier cities. 89% of the members are highly educated people with bachelor's degree or above, and loyal users account for 98%.

In the above, we have demonstrated the business value and importance of paid members. Now we can start designing the rights and interests of paid members and the prices of paid members. However, before starting the planning of rights and interests, we have to think clearly whether the types should be subdivided into paid members and whether they should be divided into levels.

For example, Tencent Video Members are divided into two categories, namely video Members and sports Members, and then video Members are divided into video VIP and video SVIP, and sports members are divided into sports VIP and sports super VIP, which is truly divided into members to the extreme.

This is not over yet. Tencent Video VIP has divided 24 VIP-level rights into 8 levels. If you want to unlock all rights and functions, you need to upgrade to the highest level V8. The V3 shown in the figure below has only 14 rights.

Tencent Video VIP has also planned the level upgrade route for you to upgrade paid members, as shown in the figure below (the content comes from Tencent Video official website). In comparison, paid members such as NetEase Cloud Music, WPS members, Baidu Netdisk, etc. are relatively simple. Although Baidu Netdisk VIP has also set levels in the equity, the equity items between each level are the same, but the specific equity data is different, such as the higher the level, the higher the equity data value.

In addition to extracting some functions from itself as the rights and interests of paid members, the products can also find some rights that users like outside the product system or even outside the company to package them into their own paid member rights and interests. For example, a certain financial lending APP product, in order to increase the attractiveness and purchase conversion rate of paid members, the company purchased Tencent Video members and packaged them to form its own rights and interests.

At this point, commercial product managers, business analysts and operations managers can start to sort out the rights and interests of paid members. The following principles can be followed when sorting out .

• Exhaustion principle: All functional points in this product that can be summarized as member (user) rights can be exhaustive as much as possible.
• Distinguishing principle: If the product has both free level member rights and paid member rights. The rights (function) items of the two should be separated as much as possible to avoid the allocation of some high-usage equity contents into the level member rights system, so that the attractiveness of paid members will be greatly weakened.
• Equity Focus: Do not set all rights and interests as level member rights or paid member rights. Users will fall into selection anxiety when they look at dozens or hundreds of rights and interests. Some basic functions of products, such as multiple playback of Tencent Video, if you divide them into equity content, it is a bit too much. Equity needs to be focused, and Tencent Video has only 24 paid member rights. At the same time, you should also pay attention to whether the rights and interests are attractive.
• Cost control: When planning the equity content, you also need to pay attention to cost control. For example, whether a paid member will lose money if he gives 5 XX yuan coupons per month? This requires the ROI of the payment member. If it is not easy to calculate, you can use the MVP or ABtest method to verify it.
• Simple principle: The rights and interests of paid members should be easy to understand, easy to disseminate and remember, and avoid incomprehensible situations. For example, Zhihu Yanxuan members can enjoy 800+ courses and 300+ magazines, which is easy to understand.

After setting up the rights and interests of paid members, we need to further consider the issuance or triggering mechanism of these rights, that is, how paid members receive these rights and interests. Some readers may ask, I have spent money to purchase paid members. Shouldn’t the member rights and interests of paid members be automatically distributed to my membership account?

Some member rights can indeed be automatically distributed, such as members can remove advertisements, members will automatically discount when purchasing specific products, members will automatically bring VIP logos, and members will automatically return double points, etc. Some rights that are extremely low in marginal costs or close to zero are automatically distributed or triggered, and the user experience is also the best.

However, it is not very cost-effective to automatically issue member rights. For example, the early JD plus members had a full-category coupon of 100 yuan per month (as shown in the figure below). This is a real silver discount. If it is automatically distributed to users, it is likely to be used up. But what if the user redeems by themselves? I believe that the overall redemption rate and usage rate are much lower.

If China Telecom launches paid members, there are Tencent Video's VIP members in its member rights. Obviously, China Telecom itself will not produce Tencent Video VIP members, and can only obtain Tencent Video VIP members through external procurement, and then package them into its own membership system [issuance] to paid members.

The following is the procurement announcement of China Telecom's Tencent Video VIP procurement. Suppose China Telecom has 100,000 monthly card, and if the system automatically issues it, the cost is 1.6 million yuan. However, if the member goes to the membership page of the Telecom APP to collect the membership and then redeem or use it, assuming that 20% is 20,000 people forget to redeemed, it can directly save 320,000 yuan in cost.

The 20% number is not something I am talking nonsense or making up. The proportion of a paid member product I participated in before is much higher than this. Just like many people who open Tencent Video VIP mainly for the core demands of closing advertising, high-definition picture quality, offline cache, and viewing paid resources. As for the novel book coupons, comic coupons, game gift packs, upgrade gift packs, movie viewing coupons, etc., there are a lot of member rights and interests, and each individual rights may not care about 80% of the members.

After setting up the content of paid membership rights, you need to price the members. Whether the paid membership system can succeed or meet the company's expectations is the most critical link, no doubt about it. Regarding the content of the pricing of paid membership products, the scope and fields involved are too broad and deep. Here is a brief description.

1. Cost of equity

The cost of equity will greatly affect the pricing of paid members, and the level of equity costs is also affected by marginal cost and supply capacity. For example, the cost of every 1 new membership content added to Tencent Video or 100 paid members is almost the same, because the procurement and production costs of member content are fixed (but the more paid members, the lower the allocation of procurement costs), the marginal cost of content is extremely low or even close to 0, and the corresponding supply capacity is particularly strong.

However, the Tencent Video rights in the above-mentioned financial product APP and China Telecom paid members will be 16 yuan more for every 1 new membership, and 1,600 yuan more for the additional 100. The marginal cost corresponding to these paid membership rights increases linearly, so the supply capacity is limited, so this factor must be taken into account when setting it up.

Therefore, the rights and interests in paid members of many products will be found within the product system or within the company group. For example, Tencent Video can use QQ Music, QQ Reading, QQ Comics and other membership qualifications as member rights of Tencent Video VIP, and QQ Music members can use Tencent Video VIP as their own member rights.

There are two benefits to doing this: QQ Music members use Tencent Video VIP as the rights and interests of QQ Music members, which can greatly reduce the procurement cost. Tencent Video VIP is sold to the outside world at 16 yuan/month, and the price of settlement for different internal BUs should be much lower. The target user groups of the two products overlap partially and are not competitive. The two can convert each other's members to each other. The latter gameplay has a professional name: joint members.

2. Profitability

For products with certain subsequent products or charging points, the price of paid members can be moderately reduced, and even subsidized to attract and retain more high-value users. The price of front-end products can be very low, and more payment points, namely back-end products, can be set up to increase users' purchases and thus obtain more profits.

The most typical example is JD plus, whose member price is 99 yuan/year. It contains 6 free shipping coupons per month + 95% off for shopping + double gold beans return + 100 yuan no-threshold coupons. The book value of the equity is obviously much higher than that of the member price of 99 yuan. Simply counting, the basic plus membership is just selling and losing.

But when JD Plus set this price, she did not think about making money by paying members themselves. Instead, she absorbed high-net-worth customers through this front-end product, completely dispelled their shopping worries, and then relied on the incremental value brought by the purchase of goods to make money.

3. Competitive landscape

Payment is a very important market competition strategy. As mentioned earlier, the Internet industry has difficulty in reaching its peak in traffic growth. Everyone has begun to strive for high-value and loyalty users in the stock market. Many products or companies are paid members, mainly hoping that these high-quality users will not fall into the friendly business and maintain the stability of the basic market.

When new customers are easy to obtain, paid members can price on demand, and make money by relying solely on the sales of the members themselves. When the industry competition is not fierce enough, paid members can price based on cost, and the overall break-even may have small profits. When the industry competition is fierce, paid members may pay money.

For example, Xiuxun, the head of Alibaba's 88VIP membership project, once said that Alibaba needs to pay for this 88VIP membership. For example, the annual fee for JD PLUS has been reduced from 198 yuan to 99 yuan and all shipping is free.

Next, let’s talk about the product pricing of paid members. You should also see from the picture below that most paid members have monthly card prices, quarterly card prices, half-year card prices, annual card prices and two (three) annual card prices. Later, in order to better promote purchase conversion, the above-mentioned price system has been added to the monthly price and continuous seasonal price.

At this time, the operators of the thief will use the monthly card price as the price anchor point. For example, the standard price of the monthly card of Tencent Video VIP is 30 yuan, and the quarterly card is 60 yuan, and the annual card is 198 yuan. Only a very small number of sensible users will buy the 30 yuan monthly card that looks more expensive. Most of them will choose the quarterly card or annual card, because no matter how to calculate the monthly card, it is not the most cost-effective, but users may forget that their original intention or trigger point of buying a member is to simply chase a certain drama or movie.

The operator of the thief has helped you calculate how much membership is converted to every day, further reducing your decision-making barriers. You may think it is expensive for 19 yuan, but if you only have 6 cents a day, you may think it is not that expensive. If you can't buy it for 6 cents, you will suffer losses and be deceived. 6 cents can make you happy to watch TV shows for 2 hours, so you buy it when your head is hot and you are impulsive.

For those members who are hesitant and do not make a decisive purchase, the system will recommend you a new customer coupon or a try coupon after it is identified, and the monthly card price will be 10 yuan and 30 cents a day, which will stimulate some users to purchase and convert.

The above actions only increase the penetration rate and conversion rate. Many users will not repurchase after the monthly or quarterly card expires. How can this stump our smart operation and product classmates? On the basis of existing one-time purchase, the price is several yuan cheaper than the normal one-time purchase price. Do you buy it?

Some readers may say that I buy at the price of continuous monthly subscriptions every time, and immediately cancel the automatic renewal after buying, and continue to buy at the price of continuous monthly subscriptions next time, so that I can save 10 yuan each time. There seems to be no restrictions on the product or operation staff. Don’t they know?

Can you think of this way of making money ? Can so many people in Tencent Video have never thought about it for so many years? Why didn’t they close this hole? Is it possible that this is carefully designed by others. The price calculated by VIP members according to the pricing process may be 20 yuan, but in order to make payment conversion, the standard price of VIP monthly card is set at 30 yuan, because they never thought of selling at 30 yuan, which is just an anchor or hook. A inappropriate example is like a merchant raising prices in advance and then holding discount activities on Double 11.

4. How to determine whether paid membership products are operating successfully

Whether the paid membership system is operating successfully depends on the following two data indicators: penetration rate and repurchase rate.

Permeability

That is, the percentage of paid members to total registered members. According to financial reports and public disclosure data, the number of paid members of Taobao, JD.com, Netease Yanxuan and Vipshop is shown in the figure below. The total number of Taobao registered users I can find is about 850 million+ (data disclosed in 2022). Based on this, it is estimated that the penetration rate of Taobao 88VIP is about 2500/85000=2.94%.

According to relevant data from Alibaba, the number of users with Taobao and Tmall under Alibaba's annual consumption of more than 10,000 yuan has reached 123 million. The average annual consumption of Taobao 88VIP users is 57,000 yuan. The average annual consumption of 88VIP members is no longer the same as nearly 3% of paid members contributed 18% of sales.

JD.com’s total registered users did not find the data, but according to the public data of JD.com’s 2021 financial report, it has 552 million active purchasing users in the past 12 months. Assuming that the number of registered members is consistent with Taobao and Tmall, its plus member penetration rate is about 3.5% (the penetration rate I found from other data is 3.2%). Overall, the penetration rate is higher than Taobao 88VIP.

According to JD Big Data and Nielsen survey data, Plus members are mainly young and middle- and high-income (65%) in first- and second-tier cities (59%). As of Q4 2021, the average annual consumption of JD Plus members is 10 times that of non-Plus members.

According to Vipshop's financial report, the proportion of paid member users in the second quarter of 2022 to the total number of users on the platform, that is, the penetration rate is about 6%, and the proportion of paid members contributed as much as 38% of the entire Vipshop transaction volume. Its paid membership price is 79 yuan per year for a continuous year, and enjoys 8 rights including a discount of 95% and unlimited free shipping.

The penetration rate of video and music products is much higher than that of e-commerce products. According to Tencent’s financial report data, the number of paid members in the first quarter of 2023 reached 113 million, a year-on-year decrease of 9%. Assuming that the total number of users is higher, it is calculated as 900 million (according to iResearch Qianfan data, Tencent Video’s active users in May 2023 was 427 million), then its penetration rate is about 13.7%.

According to NetEase Cloud Music’s financial report, its monthly active users in 2023 were about 206 million yuan, the monthly paid users were 44.12 million, and the active user payment rate was about 21.4%. Assuming that the total number of users is higher than that, it is considered to be 400 million, its penetration rate is about 11%.

Repurchase (renewal) rate

According to data published by JD.com in 2018, the repurchase (renewal) rate of JD.com plus is around 80%, that is, about 20 of 100 plus members have not renewed and repurchased after the membership expires. As long as the number of new people/speed of plus members is greater than the number of lost people/speed, the number of plus members will grow positively.

The figure below is the renewal rate data for a certain period of time in a paid member project I have participated in. The average initiated renewal rate in column 4 is about 85.7%, of which 14.3% of the difference is that the automatic renewal canceled before the automatic renewal expires after the customer purchases a paid member. The renewal success rate of the initiated renewal order is about 68%, and the overall renewal success rate is that the 4th column * 6th column = 7th column data, and the overall renewal (repurchase) rate is about 58%.

As can be seen from the above figure (for privacy reasons, the absolute value data in the 235th column of the above figure is uniformly multiplied by a certain coefficient), and the renewal (repurchase) rate can be divided into three indicators: initiating renewal rate, renewal success rate, and overall renewal conversion rate. The first two items will directly affect the final overall renewal conversion rate.

My previous project experience shows that monthly card members account for more than 80% of all paid members, and the remaining 2 become quarterly/year cards. Therefore, when analyzing the renewal rate, the above three conversion rate data should be counted according to the monthly card, quarterly card and annual card.

Finally, let’s talk about the operational growth strategies of paid members. Just like the math problems we often do in primary schools, we open the water inlet and the water outlet at the same time, and ask how long it takes to fill the pool. The operation and growth of the operation and growth of the members are also applicable. Increasing the water inlet corresponds to the new addition of paid members, and reducing the water outlet corresponds to the loss of paid members.

So the question is, how to increase the water inlet? Let’s take Tencent Video VIP as an example. New paid members = number of ordinary members * conversion rate. The disassembly means that as many ordinary users know that there is a member VIP, there must be enough VIP promotion and exposure, and the VIP page must be displayed silently in every detail of the product.

For example, the NetEase Cloud Music page shown in the figure below will have weak exposure and display of VIP-related copy on the music playback page and my music page. When users log in, there will be strong exposure prompts for VIP paid members. The combination of strong and weak interspers allows ordinary members to feel the existence of VIP members in the main operation path or page as much as possible.

As for which pages to display and expose VIP members, products and operators need to constantly try. However, the process of trying cannot disturb the user's main operation process, and the display and exposure of these member content must be coordinated with the overall UI of the product. If the part on the left box on the above picture is selected to create a conspicuous red, blue and green background, the prompt is eye-catching, but it destroys the overall tone of the product.

The second step is to guide ordinary members to try to purchase VIP members through copywriting and speech. For example, Zhihu Yanxuan members have thousands of good books and 1 million+ selected member rights, such as NetEase Cloud Music Vinyl VIP VIP library + 300 free downloads per month. Further attract ordinary members through more tempting and intuitive numbers.

Then give ordinary users a limited-time special price or discount. For example, Tencent Video's continuous seasonal VIP membership with an original price of 68 yuan is now only 45 yuan, or Pinduoduo is safe to guarantee that the difference will not be refunded + full refund if it is not used, or use it first and then buy it, etc., further improve the purchasing intention of ordinary users and thus increase the successful purchase rate.

To reduce the loss of paid members, the corresponding strategies are similar. For example, to allow paid members to truly feel the convenience brought by paid members and the benefits of real money during the payment period, this also requires continuous strengthening and reminding in the process of product design. For example, JD Plus member reminds users that you have saved a total of XXXX yuan since you opened the plus membership. The discount amount of plus members is listed separately during order settlement, and the returned gold beans are also indicated as a double return from plus members.

At the same time, for those members who have purchased paid members but have low overall rights or certain rights usage rates, they need to match corresponding operational strategies. For example, after I purchased Tencent Video VIP, I will receive 2 movie viewing coupons every month, but I have never used this coupon. Operators can do precise marketing to remind me that the newly launched movie "Hot Hot" can use free movie viewing coupons.

For example, the bonus movies and friends function in Tencent Video VIP rights. Only 5% of all VIP members have tried this function, which is far lower than the usage rate of other functions. Then the product manager must consider whether the entry to this function is too deep, whether the experience and guidance are not done enough, and whether the member’s operation steps are too many. Operators must think about how to increase the usage rate.

If the overall number of users and usage ratios of certain equity items in Tencent Video are relatively low among the 28 equity interests, then we have to consider whether to replace them with other new equity interests, or develop some new equity interests, such as comments and support pictures, such as VIP member comments being ranked first, etc.

Of course, the most core thing is that VIP members should have as many video resources and content as possible. In the early days, I was a VIP member of Youku Video, but the video content of Youku was much less than iQiyi and Tencent Videos, so I basically did not continue Youku's VIP later.

6. Level rights

Going back to the relationship diagram at the beginning of the article, some products may have both paid membership rights and level membership rights, namely module 5 and module 6 shown in the figure below, such as most e-commerce transaction products such as JD.com Taobao Vipshop, etc. Some products may not be suitable for paid members, so they will only have level membership rights, such as Alipay’s membership rights.

As the name suggests, level rights are linked to the user's level. The higher the user's level, the richer the corresponding level rights content will be. The series of rules and rules for level member rights are similar to the principles of paid member rights, so I will not repeat them here.

For products that have both paid member rights and level member rights, when setting up equity content, you must pay attention to making corresponding distinctions. However, you cannot tilt all rights to paid members because of the three highs of paid members (high consumption, high activity, and high loyalty). Whether it is JD.com or Taobao, the proportion of ordinary members is more than 95%. You cannot fool them when designing level member rights.

The previous article has spent a lot of space to introduce paid membership rights. We cannot think that these users are not as good as those of paid users just because they are all aimed at ordinary users, and then they are not concerned about the product design and operation of the level membership rights. If you think so, the road will be narrow.

Ordinary users do not buy paid members because they do not belong to the [ Three Highs ] people. It is possible that some users adhere to the consumption concept of [Bike a bicycle and go to the bar, save flowers] and think there is no need to buy it. Some users think that the premium rights that have not impressed the paid members are not worth buying, and some people think that the value of paid members has decreased and no renewal is not renewed (for example, I have not renewed Youku members). Of course, there are many reasons, but the final result is that ordinary users account for the vast majority.

After product managers and operation personnel have prepared the level membership rights of ordinary members, these users can also contribute value to the platform, such as JD.com's ordinary members. Of course, some people will say that the proportion of ordinary members of Tencent Video accounts for as much as 80%. They will only consume Tencent Video's traffic and bandwidth, but they contribute active users to Tencent Video (the financial report will be better). They watched the advertisements at the beginning and middle of the video, and contributed advertising fees to the platform.

At this point, the entire content of the membership system has been basically finished. Due to space and energy, some chapters have been briefly introduced and cannot be fully developed here. I hope readers will forgive me.