As the weak in this game, consumers are constantly teased and guided by these true and false price wars. However, in today's shopping malls, there is another kind of enterprise that won the business war not through a simple and rude price war, but through the full use and mining of data. The most typical example is Amazon (Amazon.com), the founder of global e-commerce. Since 1995 first sold books online, Amazon has completely subverted the market rules and competitive relations of many industries starting from the book industry with lightning speed. In 10 years, many century-old shops like Borders and Barnes and Noble were forced to go bankrupt or on the verge of bankruptcy. The fundamental reason for Amazon's success in the less profitable book industry competition lies in its strategic understanding and application of data. When people still don't quite understand what e-commerce is, Amazon has obtained unprecedented rich user behavior information that traditional stores can't match through the Internet, and conducted in-depth analysis and mining. What is "user behavior information"? Simply put, it refers to all the behaviors of users on the website, such as searching, browsing, rating, commenting, adding shopping baskets, taking out shopping baskets, adding wish lists, purchasing, using discount coupons, and returning goods. Even related behaviors on third-party websites, such as comparing prices, watching related reviews, participating in discussions, communicating on social media, interacting with friends, etc. Compared with the information related to the final transaction that stores can usually collect, such as purchases, returns, discounts, coupons, etc. The outstanding feature of e-commerce is that it can collect a lot of customer behavior information before buying, rather than the transaction information collected by stores. In the field of e-commerce, the amount of user behavior information is unimaginable. According to the incomplete statistics of companies focusing on user behavior analysis in e-commerce industry, a user has to browse 5 websites and 36 pages on average before choosing a product, and has dozens of interactions on social media and search engines. If all the collected data are integrated and deduced, a user's purchase may be affected by thousands of behavioral dimensions. For a medium-sized e-commerce company with a daily average of nearly one million PU, this represents the active data of nearly 1tb a day. From the perspective of the whole China e-commerce, it means thousands of TB of active data every day. It is these pre-purchase behavior information that can profoundly reflect the purchase psychology and purchase intention of potential customers. For example, customer A browsed five TV sets in succession, including four domestic brands S and 1 foreign brands T; 4 models are LED technology, and 1 model is LCD technology; The five prices are 4,599 yuan, 5 199 yuan, 5,499 yuan, 5,999 yuan and 7,999 yuan respectively; These behaviors reflect the brand recognition and tendency of customer A to some extent, such as favoring domestic brands and medium-priced LED TVs. Customer B browsed 6 TV sets continuously, including 2 foreign brands T, 2 foreign brands V and 2 domestic brands S; 4 models are LED technology and 2 models are LCD technology; The six prices are 5999 yuan, 7999 yuan, 8300 yuan, 9200 yuan, 9999 yuan, 1 1050 yuan respectively; Similarly, these behaviors also reflect the brand recognition and tendency of customer B to some extent, such as favoring imported brands and high-priced LED TVs. Through the analysis and understanding of these behavioral information, Amazon makes intimate service and personalized recommendation to customers. For example, when a customer browses a variety of TV sets without making a purchase, within a certain period of time, he actively sends the customer promotional information of another TV set suitable for the customer's brand, price and model by e-mail; For another example, when customers go back to the website to browse the refrigerator again, they can recommend domestic medium-priced refrigerators to customer A and imported high-priced goods to customer B. This personalized recommendation service is often very effective, which can not only improve the customer's purchase intention, shorten the purchase path and time, but also capture the customer's best purchase impulse at a more appropriate time, reduce the unreasonable harassment of traditional marketing methods to customers and enhance the user experience. It is a good means to achieve multiple goals. Looking at the successful e-commerce companies at home and abroad, they have invested a lot of money in the analysis and utilization of user behavior information. Their high understanding and application of data strategy is worth learning and learning from domestic e-commerce.