JD.COM shared: the new understanding and application of enterprise big data
Big data has been closely linked with our daily life.
Give an example of a scene casually. For example, when I woke up in the morning, I found that the quality of my sleep last night was not very good through the data of my smart watch. After washing my face and brushing my teeth in the morning, I walked more than 1, steps to Liudaokou subway to swipe my card and take the subway. I arrived at the Olympic Park for 3 yuan at two stops. On the subway, I found that the price of a pair of Nike basketball shoes I had browsed before was reduced through the mobile phone client in JD.COM. JD.COM took the initiative to push this product information, and I immediately placed an order to buy it, saving money.
in this process, I personally produced sleep data, walking distance data, subway credit card consumption data, subway starting point and destination geographic data, JD.COM shopping data and WeChat friends circle data, so as a big data producer, I produced so much data at once. As a big data consumer, when I browse JD.COM Mall or app in the future, the system may recommend me pillows, basketball shoes or other products related to basketball shoes to improve my sleep intelligence, and friends in my circle of friends may also buy them because of my sharing.
The data we produce, after being obtained by enterprises, especially internet companies, are clustered, split and predicted by mathematical statistics and mining algorithms to get more relevant data, and each of us is described as a label through these data. Such as gender, marital status, hobbies, income, whether you like sports, promotion sensitivity, etc., so we get many attributes of each of us, such as basic population attributes, purchasing power, behavioral characteristics, social networks, psychological characteristics, hobbies and so on.
after enterprises have mastered these data, how can they use them? Is it through these data to do marketing, such as accurate marketing, accurate advertising, and accurate recommendation of goods? Or refine the internal operation management of the enterprise through these data? Or use these data to improve the production process and guide the secondary research and development of products? It depends on the level of corporate big data practice. The good application of big data can really be upgraded to a strategic height, and it is not used well. Big data is the icing on the cake and dispensable.
according to the clustering thinking of data mining, enterprise data can be divided into internal data and external data, and internal data can be simply divided into financial data and supply chain data (the concept of large supply chain). Of course, there are many differences in the business contents of enterprises in different industries. For example, in the financial industry, there may be more financial aspects such as investment, financing and cash management, and less supply chain, while in the manufacturing or circulation service industry, there will be more data related to supply chain.
financial data are mainly financial statements, especially the three major financial statements, namely balance sheet, income statement and cash flow statement. Then there is the general ledger. Accounting in the general ledger will involve subjects, and we will also set up supplementary accounting if the subjects are not enough. Most enterprises also make budgets every year, and most of the budgets are also formulated around financial indicators, or the business budgets are reversed based on financial budgets. Of course, one of the big pieces of financial management is fund management.
there will be more kinds of data in the supply chain, from the suppliers in the upstream of the supply chain to the consumers in the downstream, including purchasing, warehousing, logistics, production, sales, after-sales and other data. Of course, we can still further refine each link.
In addition, I believe that no company is engaged in production and marketing behind closed doors, and it should actively refer to external data, including national policies, economic environment, stock market conditions, competitors and prices of major raw materials. Overall architecture of big data
Most enterprises should implement BI system or report automation system. If these systems are planned and constructed by Party B, the system scheme architecture diagram they made during the planning or implementation process is nothing more than three levels and at most four levels.
from bottom to top, the first level of metadata layer or data source layer is the data, finance, supply chain, human resources, budget and so on of our business application system.
the second level is called big data storage layer, which is to collect the data sources of each level below into a data warehouse, and then go to the third level, the analysis model layer, and build an analysis model based on the data warehouse. Some schemes even omit the analysis model layer directly and go directly to the data display layer of the last level to display the data in the analysis model. According to the author's many years of experience, such an organization form can be called BI system at most, but it can't be called big data system.
JD.COM Big Data is not a single system or product, and JD.COM Big Data application has been integrated into every business application system. Our big data collection platform automatically collects all data on Hadoop platform regularly and in real time without affecting the efficiency of the system or product and the customer experience. With the big data platform as the core, the results after processing, processing, analysis and mining are distributed to various business systems and data products, such as shopping malls, sales, data compasses and navigation. The following figure is for reference only: enterprise big data application level
Not every enterprise is a JD.COM, nor is it an Internet company, and not every enterprise's business needs the support of big data. Under the premise of meeting their own business needs, can enterprises also play with small data applications? The answer is yes, big data applications can also be hierarchical, and each level meets the needs of enterprises for different levels of data. It is roughly divided into five levels, and each level is progressive step by step.
1. business monitoring
this is the primary stage of big data application, that is, the traditional DW/BI stage. At this stage, enterprises deploy business intelligence (BI) solution, which is actually an automated reporting system to monitor the operation of existing businesses.
business monitoring, sometimes called Business Performance Management, means that an enterprise uses basic analysis methods to warn that the business operation is lower than or higher than expected, and automatically sends relevant warning information to the corresponding business and management personnel. Enterprise business and management personnel can grasp the business operation in advance according to the early warning rules formulated before, realize early warning, and help them to take some measures and means with pertinence and foresight to prevent it before it happens.
There are two key points at this stage. One is the design of early warning rules. The commonly used methods include reference methods (comparison of the same period, similar marketing activities, benchmarking in the same industry) or indicator methods (brand development, customer satisfaction, product performance, financial analysis). Indicator analysis is to select reasonable indicators. Of course, the selection of reasonable indicators here is easier said than done, but actually it takes a lot of brains to do. Let me give you an example I have encountered before. At that time, it was a scheme design for a discrete manufacturing enterprise. A very important indicator of their performance evaluation in inventory management was inventory turnover rate or inventory turnover days, which was originally a very normal and frequently used indicator. However, there were false stock-out and false stock-in in in this unit's inventory management, which caused the inventory turnover rate to look very good. Later, after considering using the moving-to-sales ratio and the stock-to-sales ratio as indicators, we combined inventory indicators and sales indicators to avoid it. The purpose of this example is to show that when we do business monitoring, it is very important to choose indicators, which not only accurately and fairly reflect the operation of this business, but also avoid artificial fraud.
2. business insight
business insight means that the system does not only provide data reports, but also "smart" reports or "smart" dashboards, so it is necessary to further predict and dig out some data that we have not known through the previous multidimensional analysis based on historical data.
For example, when I was doing a project for a hotel chain in Hangzhou, we needed to make something more interesting based on the operating data of the hotels that the hotel had invested in nationwide. For example, we needed to predict the return on investment and payback period of a newly invested hotel according to the investment in the renovation of the previously invested hotels, the current occupancy rate of different grades, the attendance rate and turnover rate of the hotel catering department, operating income, costs and competitors' hotels in local cities. In addition, there is the DuPont analysis that is often used in financial analysis. Simply speaking, DuPont analysis is a model that comprehensively analyzes the financial performance of the whole enterprise from the financial point of view. Its basic principle is that the top is ROE. For ROE, we can decompose it into ROA× equity multiplier, and ROA can be divided into net sales rate× asset turnover rate, and then it is decomposed again, and finally it becomes a tree structure full of financial indicators. Because these financial indicators are all calculated through financial statement items, accounting subjects and auxiliary accounting, there is a very urgent logical relationship between them. In this way, we can calculate some technical means to realize simulation prediction. For example, when making the budget or planning for next year, we want to adjust some financial indicators in advance, and other related indicators will also be linked, such as increasing net profit by 1%, and what other indicators need to achieve. This can help us to predict in advance and make better plans and budgets.
Of course, there are many things that can be predicted at this stage. For example, in the retail industry, the sales of most categories have a sales cycle, and we can predict the sales based on the sales cycle. It is also possible to accurately target the target group for targeted marketing according to the relationship between historical users' response to different marketing methods, marketing expenses, marketing goods and marketing effects, so as to improve marketing efficiency and reduce marketing costs.
3. business optimization
business optimization is still very attractive to most enterprises, which is also the goal that many enterprises think about day and night. In fact, at this stage, we can do it step by step, at least the enterprise has the ability to embed analytical technology into business operations. Here is an example of a case that we have done for a traditional enterprise before. Like most enterprises, this enterprise also has an ERP system. In the procurement process, we can introduce the supplier performance model. Of course, there may be many factors to be considered in this supplier performance model, such as supply quality, supply efficiency, defective product rate, after-sales service and many other factors. When purchasing, the purchaser can choose the appropriate supplier independently according to the supplier performance model. This is an example. In addition, the market price of main raw materials can be connected to the procurement interface in real time, so that procurement managers can master the procurement cycle and arrange the procurement plan reasonably.
In the retail industry, we all know that there is a strong correlation between commodities, users and commodities, just like the examples of beer and diapers, chocolate and condoms. Here, you can talk a little bit about how most e-commerce companies do it. We can find out the relationship between every two products through these products in the records of purchase. This relationship is not equal. For example, users who buy mobile phones generally buy mobile phone cases at the same time, and people who buy mobile phone cases do not necessarily buy mobile phones. This shows that there is a relationship between mobile phones and mobile phone cases, and it is a strong relationship. The relationship between the mobile phone case and the mobile phone is weak, and the strength of the relationship here is explained by the coefficient. So this relationship between commodities and commodities, we form a commodity model. Based on this product model, we can better recommend the products that users have browsed, purchased, collected and commented. After talking about goods, let's talk about users. Users can find the relationship between users through similar browsing behavior, search behavior, comment behavior and purchase behavior. Based on the behavioral relationship between users, we can recommend some products that other users who are highly related to him buy or are interested in. This is also the common practice of many Internet companies in advertising recommendation, commodity recommendation and promotion information recommendation.
4. data profitability
data profitability means that we often talk about data realization, and one way of data profitability is data productization. At present, there are many data service companies that can collect mobile games, app usage, user behavior and other data, and realize the purpose of realizing cash through their data mining and analysis technology and then output through the behavior of products or services. In addition, mobile phone manufacturers, such as Xiaomi and Huawei, all have hundreds of millions of active users and master the behavior data of first-hand users in mobile phones, even including payment data. There are many aspects that can be realized, and what limits them is their ideas. In addition, more and more traditional manufacturers have digitized their products. For example, cars+big data have become Tesla, and homes+big data have become smart homes. Of course, there are many examples here.
5. Business reshaping
Business reshaping should be the highest stage of the big data maturity model. At this stage, some enterprises hope to transform their business models into new services in new markets by analyzing customer usage patterns, product performance behaviors and overall market trends, such as JD.COM's new business, JD Finance and JD Intelligence. In addition, we can use our imagination. What businesses of BAT are developed based on the main business data? Can we think of many?
There are not many enterprises in China and even the world that really have big data. We are lucky to have big data in the whole value chain of e-commerce. How can we tap this gold mine? Only our own ideas limit us.
The above is the related content shared by Xiaobian about the new understanding and application of JD.COM's sharing of corporate big data. For more information, you can pay attention to the sharing of more dry goods by Global Ivy League.