4 most commonly used user segmentation methods!

Before writing today's notes, let's review the APP's push path management , authority and push channel management , push time management , as well as precise push and push content strategy , as well as post-PUSH data detection and "public relations" in terms of "content" .

In the second sharing session, the editor wrote about the "methodology" of user awakening , how to recall users on a large scale , and how to adopt mechanism-based "awakening" and "recalling users" . In the first sharing session, the initial construction of large-scale user operations was written. For some specific details, please refer to the serialized notes, which are divided around operations. Today, the focus will be on sharing "4 commonly used stratification methods in the construction of user stratification system and the essence of user operation" .

1 - "User stratification and segmentation" -

The essence of user segmentation is a refined operation that segments users based on user characteristics, user behavior, etc. I wrote about large-scale user operations in the first series, and the strategy is actually refined operations . Why is it necessary to carry out user segmentation operations in a wide range of Internet products? In fact, it is very easy to understand. The reason is that in Internet products, users have various differences .

For example, users may have different roles and be divided into B-end users and C-end users, into contributing content and consuming content, PGC and UGC. In e-commerce products, there may also be a distinction between merchants and consumers, and so on. Another situation is that there are differences in user behavior on an APP, such as registered users, paying users, active users, downloading users, etc.

Differences in user characteristics lead to the personalization of user demands and needs. In the process of operation, we cannot use one set of solutions to meet the needs of all users, so we have to produce "fine-grained operations" to "layer" to meet the needs of different users. This is the essence of user stratification.

Let me give you an example:

For example, "Maoyan Movie APP"

In the Maoyan Movie APP, when we open it, we will see a lot of different UGC content. These UGC contents are targeted at different user roles. For example, users may think that there may be some celebrities, such as director Stephen Chow, who will do some movie reviews, and there may also be some "movie experts", "reporters" and "movie editors" who will express different opinions on the movies. Finally, there are UGC posted by ordinary users, such as their viewing experience.

Then we may find that they are all UGC on the platform, but the users have different roles on the platform. For example, there are content consuming users, high-quality UGC users, entertainment critics, celebrities, etc. What are the demands of each person? What kind of roles do we want users to play in the APP? What kind of differences will there be , including what kind of value we need to give to different users, and distinguish these users. This is the purpose of stratification. Toutiao products also follow this model.

I'm giving an example:

About "User Tiered Operations of Maoyan Movies"

Maoyan Movie divides users into three levels through another dimension: ordinary users, movie lovers, and movie enthusiasts. The characteristics of each type of user are different. For example, movie lovers are users who focus on cinema films and have the motivation and ability to contribute UGC. Movie enthusiasts are not limited to theater films, have a wide coverage of movies and are discerning. Then, for each type of user, their operational measures are different around the behavior of being active or buying tickets, such as different PUSH content and different planned activities.

Through this case, we may find that in an APP, user segmentation can be carried out from multiple angles, and then targeted and refined operations can be carried out. The early stage of user stratification operation is to carry out refined user segmentation , effectively segment users, and then implement targeted strategies to achieve more efficient and refined operations.

So how do we implement refined “user segmentation”?

The simplest method of user segmentation is "centered on the user's growth path"

For example , for a paid product APP, the process from user registration to purchase is called a path. We can divide the user according to the stage and link they are in . This is one of the simplest user segmentations, as follows:

Purchase - Paying User

Place an order - interested users

Usage - Active Users

Download - Download User

Register - Register User

The above stratification is the most basic. In fact, by applying the above formula, we can make "nested" stratification. For example, among the users who pay to purchase, we can divide the users into A, B, and C. Among the interested users who place orders, we can divide the users into interested users A, B, C, etc.

User stratification and grouping:

On the Internet, another interpretation of user stratification is called user "grouping" . The "grouping" here does not mean to classify users into communities, but to "divide them based on user attributes." Because a user has multiple attributes. For example, "I differentiate user segments based on user value" may exist, including those who have paid and those who have not paid.

Clustering means, for example, a user has multiple labels . I want to find users who "like to watch movies on weekends", and I want to find users in the "female group" who have their periods "on the 10th to 15th of every month". These are "labels", and among these users, I am looking for "members" and "non-members", which is "clustering".

Suppose I want to perform stratification in the initial stage, from purchase, ordering, use, downloading, registration above, a user can only be in one stage, multiple labels, such a logic. For example, I mentioned group A, group B, and group C above. This is more inclined to "grouping", that is, "stratification and grouping".

2 - "The 4 most common types of stratification" -

Not all products require the AARRR model. Different types of Internet products have different stratification methods. The author has compiled four commonly used formula models for user stratification in four dimensions . These methods need to be applied according to different product forms, namely:

1. The main business chain has a high degree of standardization

Let’s take an example, such as the tool category “alarm clock”. You will find that for users, the process and links of using the product “alarm clock” are standard. It is nothing more than setting a time first, and when the time comes, the "alarm clock" will ring, and then we can cancel it. The process is very simple and standardized , which means that the main business chain has a high degree of standardization.

2. Low standardization of the main business chain

Let’s take an example to analyze, such as “reading apps” and “educational products”. For these types of products, first, “the population is from different regions”, second, “the population is at different ages”, and third, “the population is at different stages”. This type of product belongs to the category of “more complex processes” and “diversified user needs”, that is, the product form is “the main chain has a low degree of standardization”, and the chain behind it is very long, very diverse, and very complicated, with many links.

3. Is it low or high that users will influence each other in a product?

For example: in some apps, users will influence each other in the app, but in others, they will not. For example, for products like "Toutiao", the possibility of cross-influence between users is relatively low. However, for products like "Zhihu" and "Weibo", the possibility of mutual influence between users is relatively high.

Once we have the above four models to determine which position a product belongs to, we can stratify it in the four most commonly used ways. The first common way is to segment and stratify according to the user's personalized characteristics. This method may belong to the second category. User needs are diverse and the chain is relatively complex. E-commerce products usually use this method.

The second common method is to segment and layer according to user identity. This method can be used in the third model above, that is, users have a high degree of influence on each other in an APP product. What does it mean? For example, in a product, user identity information is visible to each other, such as Weibo, and users can influence each other, such as by adding V to users, giving users some honors, or medals, that is, one group of people can influence another group of people in the APP.

User value segmentation and AARRR model stratification

There are two other stratification methods, which are basically suitable for all products, namely user value segmentation stratification and AARRR model stratification . "User value segmentation and stratification" means that we judge whether the value of user contribution is high or low through a certain dimension of the APP , such as "high contribution in the workplace" and "high contribution in the technology field". This is a typical value segmentation and stratification.

The AARRR model is a model we often see in a growth environment. Through this model, we can also complete a rough stratification of users and make some targeted operation strategies. It can also be defined as a growth model, or understood as a layered model.

Summarize:

To sum up, the main content shared today is divided into two units and six small sections , namely " The difference and essence of user stratification and grouping", which includes the definition of stratification and the definition of grouping . The difference between stratification and grouping is that stratification is usually defined based on a user's "download", "registration", "payment" and other situations, while grouping is defined based on the different interest tag dimensions of a user. The six sub-sections include personalized trait segmentation , identity segmentation , value segmentation , and AARRR model segmentation.

Author: Bu Jingyun

Source: Drainage Xiabu Jingyun