In essence, crowd tags depict the characteristics of the crowd, such as the height and thinness we often say. The crowd tag used by e-commerce is of course not only tall and thin, but also includes some other characteristics. It can be said that a good crowd labeling system is the key to the accurate marketing and recommendation system of e-commerce platform. If there were no crowd labels, you wouldn't see the recommended goods we saw when shopping.

Common crowd tags include three types: basic attribute crowd tag, behavior attribute crowd tag and attitude attribute crowd tag. Basic Attribute Crowd Tags are basic crowd attributes, such as gender, age, height, weight, income, marital status and other relatively fixed characters. Behavior attribute crowd tag refers to the behavior attributes you generate in the e-commerce network, including characteristics such as residence time, bounce rate, conversion rate, number of pages visited, and access interval. In addition, the attitude attribute crowd label refers to the attitude attribute, such as whether you are interested in a certain product.

Based on such crowd tags, we can subdivide the crowd, cluster hundreds of millions of visitors, recommend products of interest according to different categories of people, improve the conversion rate of our products, and realize up-selling or cross-selling. Of course, if your data processing ability is sufficient, you can also directly tailor a data scheme for everyone to realize thousands of user scenarios.

Unfortunately, however, the label system is not as easy to build as expected, and it is often subject to the business level and modeling ability of the data builder. Of course, in this respect, Tmall and JD.COM are the best e-commerce platforms.

It should be noted that sometimes we can't perfectly summarize people's characteristic attributes with static attribute tags. For example, if you are a 23-year-old male, will you like football supplies? That's not true. Only when you click on the relevant products around football can you be sure that you are interested in football, and then send you the relevant information about football.