Not all behavioral data are valuable. For e-commerce, its main demand for big data analysis can be reflected in two aspects: one is to quickly reflect problems, and the other is to discover new user groups.
For the latter, which has attracted much attention, e-commerce hopes to analyze existing data through intelligent networking, discover and predict users' interests, stimulate users' purchasing enthusiasm, and push products to specific groups.
At present, the common way in the industry is to guess preferences through historical information and records left by users on the network, such as the recommendation of related books, air tickets and flights. However, the miscalculation may lie in the unsatisfactory accuracy and recommendation timing. For example, the system is still recommending round-trip air tickets after the user has returned from a trip.
At present, there is a research direction in the United States, which analyzes the personalized dimension of users through unstructured data analysis technology, including analyzing the personal status information updated by users on the Internet, such as Twitter and Facebook, and inferring the personality and characteristics of users, so as to accurately define individuals, realize labeling, and feed back to merchants to match with users in the target market, thus realizing product association.
In this regard, Dr. Derek Wang (Wang Xiaoyu), an American data analysis scientist, the founder of Taste Analytics and the five major visualization research centers in the United States, said that the traditional way needs to analyze based on a large amount of behavioral data, and all actions are valuable, but this is not the case, which easily leads to unsatisfactory grasp of accuracy and timing; Through the personalized dimension analysis of people's real language, speaking style and evaluation content left on the internet, it is closer to people's real nature, including of course buying preferences. Only in this way can we achieve more accurate product purchase demand mining.
The "welfare" of e-commerce merchants
At present, this analysis technology can directly release its effectiveness on the e-commerce platform, which is a solution for small and medium-sized merchants: analyzing users' evaluation of products to optimize products and enhance user experience.
For example, Derek Wang said that through the data analysis platform of Taste Analytics Signals, headset merchants on Amazon platform can semantically analyze users' product evaluation on the platform and messages on Facebook, and get users' feedback on headset brand, battery life, variety and model, as well as product analysis between different products such as Bose and Sony.
This is undoubtedly very useful for a large number of Amazon, Newegg and Yi Bei merchants in the United States, and can optimize products and sales processes in time.
Another typical application is the e-commerce platform itself. A famous large-scale household sales enterprise in the United States buys and sells products by stimulating network traffic on its e-commerce network platform. Using the data analysis platform, we not only found and solved the problem of users swiping their cards twice, but also observed the uneven distribution of network traffic in a week, and then changed the marketing process through marketing promotion.
(Analysis results of a popular soda in Amazon with the platform of Taste Analytics Signals)
Decisions are made based on data, not the data itself.
The characteristics of users come from text analysis, and every sentence users say on the internet is likely to become an analysis point. Undoubtedly, more data will effectively match user behavior and improve the accuracy of analysis, and social platforms provide a good source of unstructured data in this respect.
In fact, American e-commerce itself has begun to integrate the data information of social networks. For example, Myhabit, a flash shopping website, advises users to log in through Amazon accounts. E-commerce Macys needs to log in with Facebook account (such integration is not uncommon in China). For users, this login method is more convenient and faster; For merchants, personal information can be associated; For big data technology/service providers, data analysis services can be expanded and deep data mining can be carried out.
In Derek Wang's view, this unstructured data analysis platform service around people can not only improve the accuracy of the results, but more importantly, it is not a recommendation system, but a process of enhancing wisdom. After all, data analysis based only on existing behaviors will lead to possible failure. From the above-mentioned air ticket recommendation to the danger of adopting mathematical models in the financial field, this has been exposed in the subprime mortgage crisis.
"The data content extracted by the machine is presented to the enterprise decision makers through images, and the decision makers make decisions after interacting with the machine. The data analysis platform is a tool to assist enterprise decision makers and play its value. " Derek Wang said.
As it happens, Steve Lor, a senior contributor in The New York Times, once commented on this when writing a book about big data. Although it is the general trend for decision-making activities to rely more and more on data and analysis, common sense should also be exerted at the same time. Experience and intuition still have a place in decision-making, and good intuition is often based on a large number of data analysis.
It is better for machines and people to work together. What's more worth mentioning is that the visual presentation of images enables internal analysts of e-commerce and businesses to easily grasp product trends even without IT background, thus helping them win the market.
Big data is really beneficial, but not all companies can successfully mine big data; Only those companies that have foresight, pay attention to the system and dare to invest will gain something. For the retail industry, there are three important strategies to help e-commerce successfully use big data.
Correctly understand big data
Never mind what big data is, it is unwise to try to calculate how much data is big data. First of all, there is no exact number or order of magnitude that can be used as the dividing line of data volume, because big data is not "quantity" but "all". Through the analysis of comprehensive data, we can find the corresponding trend and further predict the future. If you want to master big data, you must have the thinking mode of "big data", that is, focus on the data that has helped complete a task. Looking for laws from huge historical data and predicting the future; Or find out the relevant factors, improve the system of searching for the best data, and get the correct data to get the maximum benefit.
How to get big data?
The popularity of big data can't be separated from the huge commercial value that Big Mac companies get from it, but it doesn't mean that big data is a "unique doll" that only big companies can afford. Small companies can also have their own "big data". Although most e-commerce companies are still in their infancy, they can also collect data, tap outstanding talents and help make more informed decisions. Data analysis can start with small data, with immediate results, and then develop into big data. Even a small coffee shop can build its own "big data" by exploring customers' drinking habits, credit card records and online positioning settings.
Although small and medium-sized enterprises do not fully have the advanced tools and models on the big data line, they can still find out the laws from their own historical data. For example, with the historical data of promotional activities for one or two months, clothing e-commerce can begin to analyze the sales performance of various categories, grasp the information of the best-selling and the most unsalable sales categories in a week or a month, and clearly understand the long-term average growth rate and compound growth rate. This data analysis method can provide the measurement index of product sales and product sales performance, so as to find out the product sales model and trend and make the next business decision. This will help enterprises achieve greater sales, and at the same time, whether there are marketing activities or not, they can monitor the sales performance of products.
Combine retail strategy with big data
From the perspective of enterprises, the greatest value of big data lies in the combination of retail strategy and big data technology. At present, the modern retail industry has become more and more complicated, because consumers have higher and higher requirements for the shopping time and shopping methods they want. Therefore, retailers need to serve customers smarter, choose the location of inventory and delivery orders more flexibly, and know more clearly how to use the collected customer data for online and offline cross-selling and up-selling. In order to achieve this goal, retailers need a customized software to formulate customer-oriented and data-based strategies in order to provide personalized services to customers.
In addition, enterprises must match the retail strategy with data analysis to the greatest extent to ensure the realization of the sales plan. One of the biggest characteristics of big data is that it can update and process information at high speed. According to this feature, once the business data is generated, corresponding strategies can be made to help enterprises gain time and space to adjust their market strategies, thus giving play to their own advantages. This is like flood warning: once there is an early warning in the upstream, the downstream should respond and adjust immediately. For example, traditional retailers who set foot on the Internet often prepare three sets of contingency strategies during the 15 minute promotion time of a batch of goods to ensure that the goods are sold as planned. By integrating retail strategy and big data, enterprises will be able to attract more consumers and provide them with customized services, thus improving product sales performance, increasing sales and expanding income.