5 user segmentation models, have you used them?

When building a user operation system, the most important thing is to establish a reasonable user stratification model. There are many different user stratification models, and different division rules can form different stratification models.

In addition to the RFM model we mentioned in the article "Finding the Most Valuable Users" and the ARGO model of intelligent user operation proposed by iResearch Ark, the most commonly used models are user loyalty and user life cycle models. In addition, in specific industry applications, we may also use the following models:

Normal distribution model

When user operation resources are very limited, we can use the normal distribution model. For example, the 80/20 rule belongs to the normal distribution form, 80% of users are near the average value of the value curve, and the remaining 20% ​​of users are the main source of profit. Generally speaking, the normal distribution model is modeled from two dimensions: profit contribution and number of users. You will find that the customers who provide the most revenue are very few. Because of user loss caused by operating costs and unexpected situations, only a few companies suffer losses, and most profits are concentrated around a constant value.

Based on the above situation, we can allocate more operating resources to focus on maintaining high-profit users, and at the same time make appropriate resource adjustments for customers who occupy most of the company's operating resources but provide limited profits. Most customers near the constant value receive standardized services, saving resources and reducing marginal costs.

Therefore, the user system is established to facilitate the optimization of different operational strategies and achieve the ultimate business indicator - profit.

Category comprehensive preference model

For e-commerce, comprehensive category preference is closely related to category; for content products, it is closely related to content type. Nowadays, the channels for obtaining content or purchasing goods are becoming more and more diversified, and users may choose different channels for different categories of goods/content. Comprehensive category preference calculates category weight through the number and frequency of users’ searches, browsing, attention, and purchases of category products, and stratifies and groups users to facilitate better user use and ultimately promote user purchases.

User activity model

User activity stratification is widely used in user operations of various websites. User activity is usually divided by PV, stay time, number of posts, etc. For e-commerce or new retail, the most important thing is purchasing behavior. Users can be divided into new users, active users, dormant customers and lost users according to certain rules based on their recent purchase frequency. Active users can be further divided into high, medium and low frequency users.

Shopping decision-making model

Shopping decision-making power is to describe and distinguish users' shopping decision-making methods through their purchasing behavior, and to group and stratify users. Users can be divided into impulsive shopping type, hesitant type, rational comparison type, etc. Shopping decision types can enhance customer service’s understanding of user psychology or improve the use of coupons, thereby increasing the final purchase rate and reducing the return and exchange rate. Understanding the proportion of different shopping groups can also be used for product design. Users can be clustered according to features such as browsing time before placing an order, number of SKUs browsed, and time from first browsing to purchase, so as to achieve personalized recommendations or push related user operation activities.

Promotion sensitive model

Common types of promotions include single product discount promotions, purchase-reduction promotions, add-on promotions, etc. Different users have different preferences for different promotion methods. After learning the promotional methods that users are willing to accept, we can conduct targeted marketing, which can effectively avoid waste of operating costs and increase the purchase conversion rate of corresponding categories and activities.