How to do “user operation”?

Let’s get down to earth and look at how “ user operation ” should be done.

“Centralized operation” for core users

User operation in the scope of Internet operation can actually be divided into two branches. One branch is the centralized operation of core users, and the other branch is the operation of large-scale user strategies. Let’s first look at the former, the centralized operation of core users.

In this type of work, there will be some characteristics and some things that we need to pay attention to. Let's look at them one by one.

1. The user scale is usually between dozens and hundreds

Why is it this scale? Because no matter how many people there are, you can’t handle it alone, and you may not be able to serve a larger number of people well. If there are too few people, it won’t be very meaningful. So usually among dozens to hundreds of people, this is the object of your operation.

2. View users from an "individual" perspective, with frequent user interactions and strong emotional relationships between users and operators, giving them maximum attention and enhancing the experience.

Because this group of people is the one you need to pay special attention to, and you also need them to bring you output, so at this time you must give them the maximum attention and enhance their experience. This includes dealing with any negative feedback we get as soon as possible, and doing everything we can to make up for any dissatisfaction they have.

3. Often have specific goals

Often, the operation for this group of specific core users should have some specific goals, which we will give examples of later.

4. It is often necessary to clarify user privileges and obligations

Around this specific goal, we often need to clarify what the privileges and obligations of this group of users are.

The above are some common points to pay attention to when conducting centralized operations on core users.

Case

Next, let’s look at two cases to help us understand this matter.

First of all, this is a recruitment post by the Zhubaijia team recruiting their founding fan team through the Douban group. The reason behind this post is that they actually want to conduct centralized operations on a group of core users, so you will see that the beginning of this post talks a lot about us recruiting fans and so on.

Then the next thing to mention is what gold medal fans do. It has a purpose. It requires gold medal fans to answer questions raised by other users in its travel community.

At the same time, it will also give gold medal fans some privileges and rewards, including free sponsored travel abroad, and often organize various offline activities or gatherings, etc., so I think this is the most basic core user operation framework.

You attract a specific group of people, hoping that they can produce some specific outputs. In order to enable them to produce specific outputs, you may give them a series of value-added services or rewards. This basically builds a very simple and clear user operation framework.

Let’s look at another case. This is a small framework that a recruitment website used to operate key corporate users. You will see that this framework is very clear and is more in line with the major principles mentioned above.

Because recruitment websites have two groups of users, one is company recruiters and the other is job seekers. There are a group of companies that are popular among people, and there is another group of companies that make you feel very low at first glance and you don’t want to pay any attention to them.

We definitely want to focus on maintaining the first category of companies on the site, so they set up such a framework at the time, hoping to maintain about hundreds of companies. What we hope these companies will do is to actively cooperate with the activities, respond as soon as possible when the activities are launched, and participate. At the same time, they must continue to provide content on the site, which is actually job positions. All positions must be updated and posted as soon as possible, and there must be a certain level of activity on the site. This activity means that when someone sends you a resume, you must deal with it as soon as possible. If you fail, you must reply to these users. This is the only way to give C-end users a better experience. We hope they do these things.

So what can we offer them? In fact, the core aspect is to provide them with more promotional resources and give them promotional positions within the site to expose them.

When there are mid- to high-end offline job fairs, participation opportunities will be actively provided to them. Various HRs may be organized in other places. Because a large number of companies are startups and their budget for employee training is not high, they will also initiate crowdfunding for course learning, and pull together several startups to crowdfund a teacher's course, and their employees can all participate. These are the services they provide to enterprises at the back end.

In a sense, they are also providing services and resources in exchange for these high-quality companies to be more active on the site, to be more cooperative with them, and to be willing to continue producing content.

So from the above two cases, you can probably feel that the direction of core user operation is almost the same as what we just saw, which is to use resources in exchange for a part of core users and what specific contributions they bring to you on the site.

However, if you want to build a community, the key thing is whether you can make the core users of your community recognize your community in the early stage, and whether your atmosphere, ideas, etc. can make everyone have a strong sense of value identification.

This requires a lot of communication. You have to convey your value proposition and achieve it through this thing. If you just give small favors or gifts, it will be difficult to help a community develop a good culture and atmosphere in the early stages.

Therefore, in the operation of core users in the early stage of the community, more emphasis may be placed on value identification and emotional interaction.

The above is the first branch, the centralized operation for core users.

“Strategic Operations” for Large Numbers of Users

After looking at the first branch, let’s focus on the second branch, which is the strategic operation for a large number of users. This also has several characteristics. Let’s look at them one by one.

1. The user scale may range from thousands to nearly a million

The first characteristic is that the user scale is obviously larger. If your user scale is too small, this is not worth it and the former approach is more suitable. So the user size could range from a few thousand to nearly a million.

2. Viewing users from a “group” perspective can clearly define group characteristics, but users have a weak sense of the official

When conducting strategic operations for a large number of users, we may need to pay more attention to data. At this time, we look at users more from a "group" perspective.

As mentioned before, when we are doing centralized operations on a certain type of users, we may need to understand each person's personal information one by one. But if you are facing a large number of users, you certainly won't be able to do that, so at this time you need to come up with some user characteristics.

You may need to extract some user characteristics such as age, places they like to go, and how they might behave on your website, etc. You may need to make some such divisions, and ultimately you need to be able to clearly define the characteristics of this group.

However, when we were doing centralized operations for core users before, users had a very strong sense of the official and would become good friends with the operations manager. But when you are doing operations for a large number of users, users may have almost no sense of the official, because you are guiding users through some designed mechanisms.

3. Design targeted operation strategies and methods for different groups

There is not much to say about this, as it has been mentioned above.

4. This consideration is usually only necessary for products with at least hundreds of thousands of users.

Finally, one particularly important thing is that only products with a user base of at least hundreds of thousands need to consider grading and classifying a large number of users, and then develop personalized operation strategies for each different type of user. If your user base is very small, it is actually completely unnecessary. It is enough for you to make some recommendations and organize some activities regularly.

In other words, your input-output ratio is not enough. If you make detailed classification of users, your input-output ratio will not be high. If your user base already has hundreds of thousands or millions, then you may be able to identify at least one category of users that has hundreds of thousands of people. At this time, if you design targeted operational strategies for them, the input-output ratio may be higher.

So these are some of its characteristics when conducting strategic operations on a large number of users.

Several common practices in strategic operations

Next, let’s look at some common practices when performing strategic operations for a large number of users.

Method 1: User Value Management

What does user value management mean? In a sense, it is necessary to make a detailed definition of the users, the products you are currently responsible for, and the entire value chain and value points. That is, what are the value points of my product for a user?

For example, registration is one point, retention is one point, willingness to remain active is one point, and willingness to pay me is one point. We need to define these points carefully, and define some key user behaviors based on the data, and then combine these user behavior nodes to intervene and guide users.

When doing strategic operations for a large number of users, you always have to deal with data. When it comes to user value management, there is actually a very classic model called the AARRR model.

This is a very typical model. As a user, if I look at it from the perspective of the product, there are probably several value nodes that a user has for the product in order. At first, they register as my users, then are willing to continue to visit, willing to stay with me for a long time, willing to renew their subscriptions with me, and finally are willing to share the things on my site.

Once you get such a model, you can build a structure to guide and intervene in users based on this model. After you have divided users into different layers according to this logic, you can actually guide and intervene in users at each layer.

The operating model of a WeChat public account

Let’s take a look at a practical example. For example, a WeChat public account is user-oriented and follows the logic just mentioned. It wants to maximize the value of a user. What is the operation model built for users?

First of all, WeChat official accounts are definitely content-driven, and then on this basis, a model is built according to the AARRR model.

First of all, in terms of acquisition, user acquisition comes from natural growth, from mutual promotion between content and other large accounts, from some activities, and from the extrapolation and dissemination of content in external channels. These are the actions taken on the first level. It can ensure that a large number of users can come in at this stage and complete the action of user acquisition.

Going up one level, at the user's active level, we will interact with users in a timely manner, including in the official account. No matter where the messages come from, it is best to reply one by one. There will also be many offline activities, and everyone can meet offline, and the feeling will be different after meeting each other.

At the same time, we will also open up some spaces for user participation, including what kind of new column to add, or what kind of new article to publish, which will be released in advance, and then interact with everyone to let everyone help make a judgment, say whether this thing is good, how it feels, and whether it is suitable for long-term development. We will often open such user participation nodes. The more users participate in your operational decisions, the stronger their perception of you will be. The last thing is content quality, so in terms of activity, we have taken some actions like these to make more users active.

The next level up is retention. In order to make users willing to stay for a long time, we focus on two things. The first thing is to optimize some experience processes. That is, after a new user follows the official account, because this is an account that focuses on content, if you want the user to stay, theoretically, after he follows you, he can form an awareness at the first time that there are so many high-quality contents and that they can be continuously produced. You need to make him aware of this, so we have made some optimizations on the automatic reply for new users. After they follow, they will be guided to have an entry to view some historical articles, with a focus on articles that convey value propositions. In this way, after reading, users will often easily form a strong perception that this place has high-quality content and the value proposition of this place is more impressive.

In addition, there was a period of time when we provided special benefits to new users. If new users followed this official account, they could download some materials or some learning-related content with some small operations. Doing so can maximize the guarantee that after a new user follows you, the first thing he does is not just a simple follow-up action, but he can also have more contact and interaction with you. After the interaction, theoretically the user retention rate will be higher.

Going one level higher, in terms of revenue, this public account offers some paid courses, so the revenue is achieved through the periodic sale of courses.

Going up one level, the core of user self-propagation is still the quality of the content. If the content is well written, users will be willing to help you spread it. This logic can be realized.

In addition, they also periodically organize some small activities and occasionally have some forwarding incentive mechanisms.

After we have sorted out the entire node around the AARRR model, you will find that in each link, we can design some operational methods to take on the value presentation of users in this link. I hope this case can also give you some inspiration.

Method 2: User lifecycle management

What kind of logic is behind this matter? It actually analyzes the massive data samples on your site, defines the typical cycle of a user from contact with the product to the final abandonment of the product, defines key nodes and key user behaviors, and makes corresponding early warnings and interventions.

For example, let’s look at an example:

From the perspective of a user's life cycle, there will probably be several nodes from the time he comes into contact with a product to the time he gradually leaves. For example, there will be registration, retention, activity, decline, churn, and recall. The whole path is roughly like this.

If we want to define some key nodes and user behaviors around this path, what might we do?

For example, Sina Weibo once discovered internally that it had 200,000 registered users within a month, of which 70,000 to 80,000 had a particularly high retention rate, while the remaining 100,000 or so users were lost one after another.

So they analyzed what happened to the 70,000 to 80,000 remaining users. Finally, through a large number of sample analyses and comparisons, we came to a conclusion that these 70,000 to 80,000 users all had one thing in common, which was that they all experienced something after registering on Weibo, and they filled in their personal information in great detail.

The retention rate of users who complete this action is very high. At this time, we can define a key user behavior. Once this key user behavior is defined, based on the logic of user operation, we should guide more users to experience this behavior. No matter what means are used, it may be that after the user fills out the form, they are given a voucher or a chance to draw a lottery, etc. These are all operational means and can be considered.

Including at the user's decline and loss nodes, we can also define some behaviors through these key comparisons.

For example, a user visits Weibo every day, but suddenly he starts to visit Weibo only once every three or four days for a period of time. This may be a sign that he is about to leave. If I define this key behavior, I can actually start to take some actions to influence him at this point, when he starts to visit Weibo once every three or four days. For example, you can select some of the more popular topics on the site every day, or push some high-quality content to him, which can extend his entire life cycle.

Therefore, from the perspective of user life cycle management, we need to define key nodes and key user behaviors.

For example, how do I define active behavior? Is logging in every day considered active or opening five or six pages on the site considered active? You can define this yourself.

The next step is to define the key user behaviors. When a behavior occurs between one node and another, the probability of entering the next link may be greater, including how long the typical time cycle may be. These things must be defined, and then early warning and intervention can be carried out at the corresponding nodes.

Method 3: User classification and classification management

The third approach is a very common and practical one, which is to grade and classify users on the site. Logically, users are divided according to various dimensions such as their contribution and activity level on the site, and user roles are defined. Usually, we should focus on maintaining core high-value users, and then influence more users through them.

Many strategic operations for a large number of users involve classifying users, but the reference dimensions of the classification may be different. Under this logic, we usually classify users according to the value they attach to the content on our site.

According to this classification, a model called user pyramid can be constructed. The top of this pyramid is often the official operators, and the next layer below is the core users. The core logic of this entire model is that as the official operations manager, I maintain a small group of core users, and then use this group of core users to help me serve these larger users. This is probably the logic.

Here, let's take a look at a real example.

This example is a user model within Meilishuo in its early days, and it is also a pyramid-shaped model. You will see that in the early days of Meilishuo, users on the site were divided into four major categories.

The top category is called fashion experts. This group of users accounts for less than 1% of all users on the site. These users are usually beautiful women with particularly good looks. Because Meilishuo is a shopping sharing community, and its main target users are female users. If it is a fashionable shopping sharing community, then you definitely need some beauties to help you hold the place together. She should be able to represent your brand image, so this is the top group of people.

Then the next level is called super experts, and this group of people accounts for about 5% of all users of the site. So what kind of people are these? This group of people are usually not that pretty, but they are particularly knowledgeable about fashion, have a strong desire to express themselves, and are capable of producing high-quality content. They will be a particularly important group to the entire Meilishuo website.

There must be users in shopping sharing communities who can produce relevant content. For example, if there is a new brand of skirt or cosmetics, someone has to try it out first and then write a bunch of things from a fashion perspective, analyzing why this color is the most popular this year, why the material of this lipstick is better, and what kind of people and body types it is suitable for, etc. After these things are written, it will be smoother for subsequent purchasing behavior. Users consume content first and then naturally make purchasing decisions, so it requires a group of super experts to continuously output high-quality content on the site.

The next level down is active users, which account for 20%-30% of the site. They may also love fashion and like to shop to dress themselves up beautifully, but compared to the previous category of super experts, they may not be able to produce high-quality content, but they are still happy to spread and share.

The bottom layer is the demand of the general public. These are a group of Type A users who come to this product to consume content and leave immediately after finding what they want.

So actually, in the early days, Meilishuo divided all site users into four categories. Theoretically, after the classification, there should be different operating strategies and methods. So what did Meilishuo do at that time?

First of all, for fashion gurus, the means of operation is to sign a contract. I will give you a high endorsement fee, but you have to listen to me. Then I will give you various announcements, and you will appear in various media or various TV programs. Anyway, you just need to actively expose yourself. In the process of exposure, I will tell everyone that you are the spokesperson of Meilishuo, and then through this process, you can help me establish the brand.

For the second type of super talents, there are basically two dimensions, namely profit sharing + popularity. Through these two dimensions, we can encourage them to continue creating content on the site. Because Meilishuo is a shopping sharing community, it actually generates income from shopping guides. It is like a super expert writes an article and links it to Taobao, and how many users place orders on Taobao. This data can be monitored. After monitoring, Taobao will give Meilishuo a portion of the shopping guide's share, and finally Meilishuo will provide a portion of the share to the super expert.

Therefore, for super talents, creating content brings real benefits. In addition, super influencers who create content on the Meilishuo site can also receive recommendations from the entire site. Meilishuo will influence active users and the general public to pay attention to these super influencers, so these super influencers can accumulate their own popularity and get some value from it.

Going down one level, for active users it’s actually not that complicated, just give them some feedback regularly. For example, regularly giving out some vouchers, regularly organizing some VIP-exclusive activities, and regularly giving out some small benefits, and then encouraging these users to be active on the site, that would be enough.

Finally, for the general public with needs below, we don’t actually need to pay special attention to them. We just need to regularly carry out some site-wide activities to allow the general public with needs to have more awareness of the high-quality content of the entire site, including some core values ​​in the products, and guide them to consume more content on the site. In this way, during the consumption process, if some users are particularly satisfied with the content on my site and have not discovered that there are so many good things on the Meilishuo site, some of them will naturally become active users.

The above is a real case. Meilishuo defines such a user pyramid model and formulates different strategies for each type of user, which ultimately helps the user ecology of the entire site form a virtuous cycle.

There are actually many things here, including regular feedback. When you first start doing it, it must be done manually, but after doing it for a while, you find that for active users, it only takes one big feedback a quarter and two or three small feedbacks. This can be done with a product mechanism, that is, the system automatically gives benefits to these active users, no longer through manual means. There are a lot of operational strategies and actions. If you find that they are relatively easy to copy, you can actually make them more mechanism-based, so this is also the awareness we need to have in our minds.

Method 4: User incentive system & growth system design

The fourth approach is actually to develop a mechanism to drive users to become more active on the site naturally. This is usually called user incentive system and user growth system design.

Typical manifestations are like points in a forum or tasks in a game. These are very obvious user incentive systems and user growth system designs. It’s quite complicated to talk about this in detail, so this is just to help everyone build a cognition and I won’t go into too much detail.

Basically, we need to design a user incentive system and a user growth system. The logic behind this is roughly the following three points:

1. Define the user behaviors that need to be incentivized, sort out available resources, and design incentive rules

For example, the user behavior that needs to be incentivized is logging in, posting, browsing, or some other behavior. This needs to be defined first. Then we need to sort out the resources that can motivate him. There may be some material resources, such as prizes, and some intangible resources, including some privileges in the product. After you complete a certain action, you can experience a certain function of mine, or after completing a certain action, you can come offline to meet us, or participate in some kind of threshold activity, etc. These are all our resources.

After you have sorted out the first two, you can then design your incentive rules, focusing on the user behaviors that you need to incentivize. It would be best if the incentive system could be aligned with the user’s growth. Of course, it doesn’t matter if it’s not aligned with the user’s growth. You can first provide some incentives and then consider how to connect them together bit by bit.

Above we have shared the four most common practices for strategic operations for a large number of users. We hope that these four practices can also enable everyone to establish a more specific understanding of user operations.

Others: Explore relevance and target your efforts

Analyze the data correlation between important indicators, adjust and optimize the user structure within the site

In addition to these four methods, there are other ideas that are still closely related to data. We can dig out some correlations from the data and take targeted operational actions. We may analyze the correlations between some important indicators to adjust or optimize the user structure within the site. In fact, this matter has also been mentioned in the previous gameplay 2, but if we want to dig deeper, we may need to look at more data.

For example, these are some key node data about the structure of Sina Weibo and user behavior.

If we mine the data and then analyze it, we can come to a conclusion that if there are more and more female users born in the 1990s on the site, it will naturally drive more active male users born in the 1980s and 1990s.

If we come to such a conclusion, then there is no doubt that we should focus our operations on how to attract more female users born in the 1990s and make this group of female users more active on the site.

The logic is actually like this: we will look at the correlation in the data. After finding this correlation, we may need to use it to adjust and optimize the user structure within the site.

So far, I have given a general introduction to the definition and specific practices of user operations. Next, I will talk about the user operation system in more detail.

Author: Xiongxiong Operation Notes

Source: Xiongxiong Operation Notes