Three major subsystems behind the large-scale user operation system

Previously, we already had a very deep understanding of user operations and large-scale user management. Next, I would like to talk to you about the three major subsystems behind the large-scale user operation system.

First, let’s go back to this picture. There are two states of large-scale operations. Beginners pay more attention to the surface and single points, while advanced players pay more attention to the core and system building. The system building part involves the sorting of user operations and the establishment of a user operation system.

In fact, the establishment of the user operation system and the sorting out of the user operation model are one and the same. Only when the user operation model is sorted out clearly can we build our user operation system on this basis. It is such a relationship.

What is the “user operation system”?

Well, we need to go deeper here and ask what exactly is a user operation system?

From the perspective of professional terminology, the word "system" actually means a complex system composed of multiple different systems. If we want to understand the user operation system , we must know what common systems (working methods) exist for large-scale user operations that can be used by us.

Because behind a system is often a more mature working method. So this brings us to what I want to share today. In the user operation system of common large-scale products, there are often three major subsystems (working methods).

3 major subsystems in large-scale user operations (working methods)

The three major subsystems basically rely on the three core guiding ideas mentioned above, namely data-driven, refined operations, and automation + mechanism-based operation.

They are user life cycle management system, user stratification (classification) operation system, and user behavior incentive system. The above three systems can be used independently and can play a significant role in improving the user value of a product.

So, next I will introduce to you what these three types of systems are, but here I will only talk about what they are. The specific application of each system is actually quite complicated, and the application methods are also very rich and diverse, so I will talk about them in detail later. Let’s take a look at them one by one.

User life cycle management system

First, the first system is called user lifecycle management system. So what is it? Simply put, it analyzes massive data samples on the site, defines the typical cycle of a user from contact with a product to final abandonment of the product, defines key nodes and key user behaviors, and makes corresponding early warnings and interventions.

It may sound a bit abstract, so let's take a look at a picture like this:

This is a very typical user life cycle model for many Internet products, especially content-based and community-based products. From the time a user first comes into contact with a product to the time he finally leaves and abandons the product, there will be several nodes in the process, including registration, retention, activity, decline, churn, recall, and then abandonment. The horizontal axis is basically time and behavior; the vertical axis is whether the user's typical value is high or low at different times.

When the data sample within a product is large enough, it can basically determine how long it takes for a user to register until he or she leaves the product. During this time period, if we want to influence some changes at each node, we may need to do some cross-analysis through the user's behavioral data and basic data to find some clues.

For example, it has been found in many community products that after a user enters the product, if the first thing he chooses to do is to complete his or her profile, he or she will enter his or her gender, age, job, location, interests and hobbies, etc. If these materials are complete, the probability of him surviving and being active in the future is much greater than if they are incomplete, so we have obtained a clue.

This includes relationship chain products like Sina Weibo, which will also discover through data analysis that if a user follows more than 5-10 other users immediately after entering the product, the possibility of his or her subsequent retention and continued activity will be greatly enhanced. This can also guide users in a targeted manner, so that they will engage in this kind of following behavior as soon as they enter the product, thereby better influencing users during the effective life cycle.

The model we see in the figure, from registration to retention to activity to decline, is actually just one type of model. When we define the user life cycle, there may be multiple types of models.

What is the process from a user's contact to their leaving? What are the key nodes in the middle? We sort it out, and then use more strategies and data analysis to explore each node to see how to influence users effectively. This is called a typical user life cycle management system.

After we find out a cycle, we will let more users follow this more typical user life cycle development roadmap to achieve more efficient user management.

User stratification (classification) operation system

Next, let’s look at the second type of system. It is actually a more modeled and systematic embodiment of the refined operations mentioned above in a product. Divide users according to their contribution, user roles, preferences, etc., define user roles, and develop targeted operation strategies for different users, as well as increase the average user value.

In this type of system, the focus is usually on maintaining core high-value users. Because Internet products often follow the 80/20 rule, 20% of our users create 80% of our value. So once you are able to find out which users are high-value, you should invest more attention and energy in them.

The user stratification operation system often presents a pyramid-shaped model , which basically divides from user managers to core users to valuable users and then to general users from top to bottom. The higher the level of the user, the higher his value is, and it is worth investing more energy. The lower the user level, the larger his base number is, but his value may often be lower. We just need to focus on enabling lower-level users to advance online and become slightly higher-level users.

Next, let’s take a look at how this pyramid model is applied in a real scenario.

Here I would like to share with you a model of user operations within the entire site in the early days of the Meilishuo product. It is also a pyramid model. In the early days, it divided the site users into four categories through user stratification and classification.

Fashion experts account for less than 1% of the total users; super experts account for about 5% of the total users; then there are active users, who account for about 20%-30% of the total users on the site; and finally there are the general public, who account for about 60%-65% of the total users.

So, how do these four types of people operate them? Let’s talk about them one by one.

First of all, fashion experts are a group of very beautiful ladies. Since Meilishuo was originally a community that focused on fashion shopping guides, this group of people carries the brand image and external publicity of my product.

Then, super influencers are a group of people who understand fashion, have the desire to express themselves, and produce high-quality content. They have a relatively strong desire to express themselves and are able to produce high-quality content. To put it simply, this group of people also pays attention to fashion and has the ability to output content, but they are not so beautiful. This is a more popular and straightforward statement. This is basically this group of people. Because Meilishuo is a shopping guide community, the community must rely on high-quality content to develop, so someone has to create content.

So, top beauties carry the brand image, and then, super influencers who are not so beautiful, they help us solve the problem of high-quality content.

Going further down, active users basically like fashion, but may not have the ability to produce high-quality content. However, they still like the content produced by the platform, so they are happy to spread and share it. This group of users is also very important and quite valuable.

Finally, the general public is basically a group of passers-by and consumers. They come to our products to consume and find content, but they will leave immediately after finding it. They will not help you share or spread it, nor will they stay to interact with you. This is basically a division of these types of users on the site.

Now that we have such a clear division, we can then carry out targeted operations for each type of user.

First, what I do with this group of fashion gurus is to sign contracts with them. After signing the contracts and paying them the money, I let them participate in various occasions to help promote my brand and establish a strong brand effect. This is a strategy for fashion gurus.

Second, for super experts, the first thing to do is to share profits, because these users will create content for us, and the content they create has the value of guiding shopping. Once they guide shopping, my site will generate income, so I can give them some profits and some material incentives. At the same time, the traffic and attention within the site are continuously input to them, allowing them to become kols and have their own influence. This group of people basically rely on profits + popularity, and have material, spiritual and virtual feedback to maintain their relationship. What we hope they can do is to continuously output content.

Third, for active users, their operations are basically very simple. We just need to regularly give them some coupons, welfare coupons, small gifts and other peripheral products and benefits to maintain their relationship and make them willing to visit the site regularly and interact with us. That will be great. I hope the value they bring is to continuously spread the high-quality content on the site.

Fourth, the last group is the general public. There is nothing much to say about this. Only when we are doing site-wide activities, perhaps some of our operational actions can cover them. When they come to us, we don't expect them to bring anything, we just want to consume content.

So this is the entire operation model. You will see that in this pyramid operation model, after we have segmented the users, we can take some clear and targeted operation actions for each level of users.

It should be mentioned that for the user stratification (classification) operation system, the pyramid user model is just one of the models we can rely on. We also have many other operation models to rely on, such as the RFM model that is often used in many e-commerce products.

As mentioned before, this part only solves the problem of cognition, not the problem of specific application. Therefore, in this part, you only need to understand what the user stratification operation system is and that there may be multiple types of models behind it. Later, I will analyze in detail what models can be used by us and how to apply these models well. This is the second type of system we are going to mention.

User behavior incentive system

The third type of system is called the user behavior incentive system. There are many ways to express user behavior incentives, and common monetization methods may include the following.

The first possibility is that in a product, users will be motivated through competition or some of their advantages. For example, through some competition charts, you can see how many people in the country you have defeated, and there are some rankings, such as WeChat Sports, etc., to show and motivate users' behaviors. This may be a common form of expression.

The second possibility is to motivate users through honor or identity, such as some level benefits, achievement medals, labels, and titles, to motivate users.

Finally, user behavior is incentivized through privileges or rights. A common example is the rights pool. For example, if you meet certain conditions in Thunder, your downloads can be accelerated. Then many products will have a points mall, allowing users to earn points during use, and the points can be redeemed. Some products are more straightforward and will directly send red envelopes, gifts, benefits, etc. to users.

These are basically manifestations of the user incentive system. A user behavior incentive system will basically sort out which user behaviors on the site need to be incentivized by me, and then I need to judge what kind of incentive method to motivate them may be more appropriate and effective. Then combine these incentive methods together to form a clear mechanism, thereby building a so-called user behavior incentive system.

Let’s take a look at an example. For example, this is Xiaohongshu’s user behavior incentive system, which is divided into several parts. First of all, it has a user points system, which is used to divide its users (Little Red Potatoes) into 11 levels. After you meet the relevant requirements of likes, comments, collections, and content production, and reach a certain score, you can upgrade.

So what can you do after upgrading? After reaching the corresponding level, users can have exclusive tags, stickers, and emoticons. This system evaluates the user's activity level and contribution to content production.

Next, in the e-commerce sector, Xiaohongshu also launched the Black Card Membership, which can be purchased for a fee. After paying for the membership, you can get a series of privileges, such as exclusive prices, limited-time purchases in advance, seven-day automatic refund with guaranteed price, etc. These are a series of corresponding rights and interests that come with the Black Card Membership. These are the two incentive systems we can see.

So what are the directions that its two incentive systems respectively hope to motivate users? Basically, it uses black cards to encourage users' consumption and activity in the e-commerce sector, and uses the Little Red Potato rating system to encourage users' content production contributions. On this basis, through these two things, users can be labeled to a certain extent and their levels can be distinguished.

After this thing is described, if you want to do user stratification operations in the future, it is possible to use it as a basis for user stratification, because in a sense, it has become a label attached to the user.

Summarize

Now that we have finished talking about the three major systems, let’s make a brief summary.

Everyone must be aware that all products with large user volumes may need to use one or more of the above methods to manage users, and the same system may perform differently in different products.

It’s like when we both use a user stratification management system, we may use the pyramid model in product A and the RFM model in product B. Therefore, the performance of a system may be different in different products. It does not mean that if all products use the same system, the final products will be the same. Everyone must understand this.

Moreover, the above three subsystems can be used independently or in combination. The independent application method should be easier to understand, but when it comes to composite applications, what might these three subsystems look like?

Possible user operation system composition in a mature product

Let’s look at another picture. For a mature product, assuming that these three subsystems are applied in this product, it may appear in the following form.

First of all, this product has a user life cycle management system, which divides users into several categories: potential users, new users, old users & loyal users, and paying users & high-value users. In terms of the division of these types of users, we may make some targeted strategies to guide more users to become paying users, high-value users, and loyal users. This is one level of application.

On this basis, a second-level subsystem was created, called the hierarchical (classified) user operation system. It means splitting the users in the site again into kol users, who may be big names or celebrities in the industry, etc., and then using some targeted means and strategies for them. The second category is expert users. These users meet certain defined conditions on the site, so we give them an identity label. These users may play a connecting role on the site, and we will also implement some targeted strategies for them. The last ones may be ordinary users who have no identity identification. For these three types of users, our superimposed operation strategies may also have some differences.

On top, there is a user behavior incentive system. We have developed a series of incentive mechanisms for user behaviors that can be motivated on the site, including points, levels, honors, guided tasks and rewards, so that users will have stronger motivation when they perform the behaviors we want them to perform on the site.

Therefore, in a product, if these three systems exist and a complex application state is produced, it is possible that this will occur. Of course, there is another possibility. For a mature and complex product, after the strategies of these subsystems have been completed, at a certain stage we may consider merging these three systems. After the merger, a simpler and fresher system may be derived. I will not elaborate on this here and will share it with you later.

The above three subsystems are introduced here. I hope everyone has a certain understanding of these three subsystems. Next, I will give an in-depth explanation of the composition, common models and applications of these three subsystems.

Author: Xiongxiong Operation Notes

Source: Xiongxiong Operation Notes