The construction and use of user Portrait 2 application

The specific applications of the user portrait include accurate pre-sales marketing, personalized recommendations for sale, and value-added services after the sale. The user's label latitude and the application are mutually Xiangcheng relations, one side can develop the application according to the existing label latitude, on the other hand can expand the dimension through the application demand, the two mutually promotes.
Our examples here are divided into 3 categories, the first is the pre-sales of precision marketing, such as e-commerce customers and business customers, need to undergo precise marketing, the outside of the site to attract users to your website. Second, the sale of personalized recommendations, is to attract these users, if the personalized way to better improve the site conversion effect. The third category is after-sales value-added services, that is, you sell the product after the end of this, and just the beginning of your contact with the user, may design to the user follow-up product consultation or spit groove, this three applications are closely related to the user portrait.

First, according to the user portrait of accurate marketing, this piece of the portal ads, Baidu's search ads or some of the current DSP companies drag and drop programmatic advertising is more familiar with, percentage of things to do is to help groups of enterprises to integrate and pull-through their own first-party data, set up enterprises to create a user portrait, to achieve omni-channel marketing.

This is what we do for a well-known enterprise customers Big data goal is to open up and build a unified consumer data platform, the creation of consumer user portrait, and based on user portrait to achieve accurate marketing. The above image is for the enterprise to build a user portrait system, with the dimensions include basic basic information, product information, financial information, risk information and asset information and so on.
And the combination of percentage of marketing butler products, can achieve trigger-type marketing, such as the user in a website to buy a mobile phone, he can immediately push the brand phone corresponding to the mobile phone accessories ads. The final effect is, through user pull-through and user portrait, to 59w potential consumers to form 4 accurate crowd by the line, is the blind click-through rate of 10 times times.




In the example of a social marketing, one of our add-on manufacturing enterprise customers, new product release period by SMS and e-mail, from the old users to find the most able to participate in the activities of fans. We use the company's CRM, customer, sales and other data, the user loyalty to the comprehensive assessment and selection of the highest loyal users as a recruitment goal, brought more than half of fans, but the cost is only 40%.

Next is the sale of personalized recommendations, which is a percentage of the creation of things to do at present, has served more than 1000 e-commerce and media customers, is the largest third-party referral service provider, mixed cent of the recommended a very big thing is to use the whole network user portrait to recommend, for example, a new application of Wangfujing Mall, In the home page to recommend, because of its history did not understand, can only recommend some popular products, but percentage point, we know that the user in other customer site behavior, such as it is very interested in cosmetics, it can recommend related cosmetics, using the whole network user portrait to solve the problem of wangfujing cold start.

This is the design architecture of the percentage recommendation engine, the core is the four building, including the scene engine, rule engine, algorithm engine and display engine, especially the rule engine is very powerful, can be more customer's business needs to be honest configuration recommendation logic, such as, push new products, clear inventory, and not only the best click-through rate.

We give 1 recommended practical application examples, one of our group buy site customers, the next network uses our recommendation engine to end his issue of the order rate, we found the site by analyzing a series of characteristics of users, such as loyalty, the regional purchase.

We adopt a series of optimization measures, for example, according to the user portrait of the variety preferences, shopping district preferences, consumption capacity preferences, such as label optimization recall, so that the click rate of group purchase site increased by 18.23%, directly under the single rate increase of 86.95%, achieved a very ideal effect.

The last example is how to combine the user portrait to provide after-sales value-added services, the above picture is our customer's application of a system, can be real-time feedback through the data interface user information, such as historical maintenance, historical consulting and so on, and to carry out knowledge recommendation, support service efficiency and satisfaction, At the same time collect customer service satisfaction data, supplement and * * Improve the user portrait information,
User portrait is a step math game,
It's a serious business problem, the best combination of business and technology.
is the best practice for implementing and data. **
The last child to the front of the content of a miss, in the big Data age, the machine to learn from the bit stream to understand the user, the construction of the user portrait becomes particularly important, is the top of the various applications of the basic building user portrait of the core is the label modeling, tag step is only a symbol, but closely related to the business, is a very good combination of business and computing points.
Some do not understand, and then come back to learn slowly.

The construction and use of user Portrait 2 application