Industry Analysis: How to use data to drive sales

Intermediary transaction SEO diagnosis Taobao guest Cloud host technology Hall

Billion State power network case Center editor-in-chief Wei

Percent Technology COO Angela Pk.

Korean Public relations Director Chen

The electric business experienced the extensive savage growth, gradually enters the fine operation stage, the data mining and the analysis also becomes the new target which the electricity merchant enterprise exerting force.

What data mining directly drives sales growth in the finer battles of this data? How do I turn data into sales?

Less than 5% power companies have data mining team

Billion power network: two to determine the current electronic commerce industry data operation to what extent? How many companies have data mining and business intelligence teams? What are the capabilities of data analysis and mining?

Chen: With the current development trend of e-commerce, data analysis for the development of enterprises, especially for the electric business is very important. The data operation makes it easy for the enterprise to analyze the user's behavior and the various business data of the competitor.

For example, in the United States, BI has become the key to E-commerce business competition. In the domestic e-commerce companies, because the data sector is difficult to create a significant short-term performance, so the data sector is easily weakened. Although some electric companies have begun to focus on data operations, the BI data is often placed in the technical Data Warehouse. If the level of the Business Intelligence department is low, and the head of the BI team is unable to meet in the management, it is difficult to analyze and excavate the data that is scattered across each business unit. In contrast, BI is an independent department in the United States, directly affiliated with the company's CEO or CFO. The Business intelligence team occupies a high position in the company.

According to our understanding, the electric business industry platform company, like Taobao and Jingdong have their own data analysis team. Taobao Cloud Platform large data analysis and calculation is very strong.

Zhang Shaofeng: From the current service point of 200 dozen Electric Company, the Electronic Business enterprise data operation degree is very low. The number of electric companies with real data mining teams is less than 5%.

In fact, if a company has a dedicated professional to do some relatively simple data statistics work, it has been calculated as a higher degree of data operation of the company. In addition, because of the low penetration rate of data mining concept, many enterprises mistakenly consider data mining as a result of some simple data statistics, so they also think their data mining level is high.

In other words, even for those companies that have a bit of data analysis capability, their level is mostly in the simple data statistics phase.

Billion Power network: Electric Business enterprise data analysis business is inclined to build their own team or outsourcing?

Chen: For E-commerce Enterprises, the data is roughly divided into four kinds: Station user data, business data, outbound user data and competition data.

For us, the station data is more intuitive and explicit, and can be tracked and analyzed by itself. and the analysis of outbound data, the need for user research or competition analysis, this part of the data source is very dispersed, more dependent on outsourcing.

For the Enterprise BI Department, this kind of outbound data analysis, including competitive data, plays a very important role in decision making. This is a difficult task, in addition to relying on the experience of the BI team, but also to the various Third-party data evaluation, take its essence to its dross.

If from the company's technical team training point of view is still not advocating outsourcing. Outsourcing words have both advantages and disadvantages, the advantage is to find a mature, with corresponding industry experience of outsourcing companies to do, the bad place is the project after the maintenance inconvenient, and may be expensive.

Zhang Shaofeng: In fact, in the strictest sense, all enterprises in the field of data mining, data analysis in this area with other companies, but a weak technology company may be the data analysis of the whole work outsourcing, and the technology is a bit more genetically strong companies may outsource part of the work. And regardless of all the current electric business enterprises are lack of technical genes, even Alibaba, Tencent these technology is very strong companies, also in the internal use of Oracle, IBM, SAS and other professional data analysis, data mining company's products or services. Some of our colleagues in the company used to work for Oracle and IBM, and they served the three internet giants.

Korea's clothing house daily PV up to 1.3 million

Billion power network: Two, it seems that the electrical business enterprise is necessary to do data analysis and mining? What is the significance of data analysis and mining to the development and planning of electric business enterprises?

Chen: When it comes to whether it's necessary to do data analysis and digging in person, first let's be clear that the application of the data should start with the problem. Enterprises have to solve the problem according to their own, around the problem to figure out how to build the decision-making framework of the enterprise. The decision framework is naturally the first to assume the solution to the problem, then look at the decisions of the management and then build the monitoring system.

The purpose of data analysis is to guide decision making direction. At present, many CEOs of enterprises, although the importance of data, but did not start from the problem, so easy to lead to data applications and business decision-making is not close.

Data analysis and mining are of great significance to the development planning of electric business enterprises.

First of all, through search engine keyword we can hold what keyword is effective, keyword can help us understand and analyze the browser's search habits and needs.

Secondly, through the website PV, UV and other data analysis, we can know the site's overall flow, as well as user experience and so on. At present, Korean clothing homes UV up to 330,000 uv,pv130 million, such a clear data on our grasp of the operation of the store has a great help.

Secondly, through the analysis of the target population, we can know some of their basic information, including age, geographical and so on, thus making more targeted design. Finally, through the browse volume and volume of data analysis we can get the basic conversion rate, and the time of the UV data statistics, more convenient for us to formulate promotional programs according to the specific circumstances.

Zhang Shaofeng: The electric business enterprise certainly has the necessity to do the data analysis and the excavation, here "personally", more is emphasizes that the electric business enterprise must participate actively, the thorough participation in this activity, but does not say all things, all tools must own independent development. Data analysis and mining system is to ensure the enterprise into the cornerstone of fine operation, without it, enterprises can not optimize operational efficiency, also can not avoid some hidden risks.

In particular, like the electrical business of this industry is characterized by agile, flat, if not based on data, do not do the corresponding data extraction, excavation, management, analysis, the equivalent of blindfolded in business, elephant General. The biggest mistake in this industry is our data use and understanding is not enough, causing headaches medical head, not from the overall macro level to grasp the entire business, if you want to be able to grasp the overall macro, it is necessary to do data mining.

Billion power network: In this May, the Korean clothing house officially online bi system, this system mainly do what aspects of data analysis? How many people are there in the current data mining Analysis Team? What aspects of the data were made by yourself? What is the use of third parties? What are the third parties currently cooperating?

Chen: At present, we have completed the Tauis system order, inventory, return and delivery data analysis chart, the new ERP system Commodity data analysis chart. The day-to-day business data can be found in an intuitive graphical way.

The BI team has now expanded to 11 people.

Billion power Q: Han clothing homes now their own data analysis team in the actual operation process encountered what problems?

Chen: Han's bi system has just been online, although the BI system has completed the visual statistics of various data, but the current analysis of the Department of Business data is still in the initial stage, so we need to solve a lot of professional problems.

Refined operation based on user

Billion power network: The current data is also a large number of electronic business analysis methods, evaluation of the main indicators of sales data what?

Chen: Mainly has the daily dispatch data analysis, the daily sales data analysis, the daily stock quantity analysis. Through these analysis charts can be intuitive to see the daily sales, order volume, cost and inventory.

The ranking from high to Low is: daily sales data analysis, daily shipping data analysis, daily inventory data analysis.

Zhang Shaofeng: Evaluation of sales data indicators are: order volume, shipments, customer unit price, total sales of goods, total sales, total gross margin, total inventory and so on. The general order of importance from high to low is: total sales or gross margin (different stages focus on different), total shipments, total inventory.

Billion power network: Korean clothing homes currently applied data analysis generated by the sales of specific data can be shared? for example, what data applications directly enhance the customer unit price?

Chen: Sales data are currently only the total, in view of the customer's CRM system has not yet been established, the customer's sales data has not been in-depth excavation. Customer Order data clustering method can be used to classify customers in the future. Provide the basis for different kinds of customers to establish different marketing strategies.

Billion Power network: If you let two-bit choose the most important or favorite data analysis and mining applications, two-bit respectively what is it?

Chen: In current data analysis and mining applications, individuals feel that the sales and delivery analysis of orders is more important. Because can intuitively see the daily sales, also directly affect the performance of various departments and the overall level of sales.

Zhang Shaofeng: There are two categories: 1 prediction class. Forecasting is a very important branch of data mining, such as predicting what users might like, and how a product will sell next month. This type of application can be applied directly to the business operation. such as personalized product recommendation engine, inventory forecasting engine and so on;

2 statistical analysis class. Mainly refers to the use of various statistical analysis methods, using a variety of graphic display means, so that managers and business operators to see some key indicators of the status quo, so that the link is worth improving.

Billion power network: If the user behavior analysis is the premise of fine operation, then the electrical business enterprises to achieve fine operation should also be from which aspects to proceed?

Chen: When it comes to fine operation, it is natural to put the user in the first place. The essence of any business process is to "meet customer expectations" and "exceed expectations". So we have to first of all to the user information for detailed analysis. Each person's interests, hobbies, personality, culture, economic situation and so on are different, in the purchase mentality will also produce differences, so formed a variety of buying motives. We believe that before the full promotion of the project, the electrical business should pay attention to the user experience, so as to enhance user loyalty.

User positioning is the basis, that data analysis is the backbone of the refinement of the application. Users in the E-commerce mall on the purchase behavior, from the potential customers into the value of the mall customers. Database will save the user's transaction information, including purchase time, goods, quantity, amount, etc., we can based on the operation of the mall data on their trading behavior analysis to estimate the value of each user, and the possibility of expanding marketing for users.

Zhang Shaofeng: E-commerce enterprises must be from the front-end marketing, midrange Web site operations to the back-end of the supply chain, logistics warehousing and other aspects of optimization to achieve fine operation.

At present, we see the vast majority of electric business enterprises in the marketing and website operation two links exist a large number of waste of resources. This year the electric business enterprise must pay more attention to the effect marketing, the station transforms and the duplicate purchase. Whether it's the front end, midrange or backend, the premise of realizing fine operation is that the electric business enterprise must establish the data-driven operation culture as well as the perfect data analysis and the business intelligence system, through the data analysis to find out the most effective marketing way, the most potential commodity, the most valuable customer group.

In addition, in order to upgrade the rate of conversion and customer repeat purchase rate, electric enterprises should also establish personalized referral system and improve customer repeat purchase rate of personalized mail marketing system.

Recommended engines for red children to contribute to the order of 15%

Billion power network: percentage point is to do the recommendation engine, is to help the enterprise analysis of users buy browsing process recommended which related products best? Is this technology still being done at home by other companies? What is the technical core?

Zhang Shaofeng: In fact, the recommendation engine does contain "association recommendation", but "Association recommendation" is not the whole of the recommendation engine, this is a common misconception that the recommendation engine is easy to fall into.

There are two application scenarios for the recommendation engine:

1. When the enterprise does not know what specific content and commodity the user specifically cares about (for example, the user just arrives at the homepage of the website or landing page, or just enter a channel page, but not to the specific article page or product page, based entirely on the user's past behavior to guess what they might like content and merchandise. This recommendation is the true meaning of the "personalized recommendation";

2, when the user has been concerned about a specific product, recommend that the product has some kind of association with other products, this recommendation is commonly referred to as "Association recommendation." Percentile Company's recommendation engine covers both types of recommendations. The percentage point is a company that specializes in Third-party recommendation engine services, with the exception of a percentage point where large companies such as Alibaba and Tencent have also used the recommended technology sporadically in some of their products.

The core of recommendation engine technology includes: User behavior modeling, Web page content modeling (including text content and image, video content), massive real-time data processing, and the ability to integrate user psychology and sociology knowledge into the recommendation engine.

The main difficulty in making a recommendation engine is: 1 algorithm. The algorithm should be advanced and stable and reliable in large data environment; 2 data. The recommendation engine is modeled entirely on the basis of real data and does not have enough high quality data to make a recommendation that has applied value

Billion power network: Many electrical business enterprises formed a number of stereotypes, it is considered that the recommendation engine adopts the similarity mining method (such as association rules and collaborative filtering), the recommendation engine "read and read, bought and bought" such recommendations through a simple database query can be completed. So what's the percentage on the recommendation engine?

Zhang Shaofeng: "Have seen also read", "bought and bought" this term, is to facilitate the consumer good understanding. In fact, this seemingly simple name of the recommendation bar, the logic behind the algorithm is not simple, more impossible through a simple database query can be completed.

As for association rules and collaborative filtering, the older algorithms that have been in use for 10-20 of years, they have many flaws, so there are a lot of other more advanced algorithms that have been used in recent years, such as a percentage point of the chief Scientist Zhou and his mentor, Professor Zhang Yicheng, who invented the mass diffusion algorithm, Thermal diffusion algorithm and so on.

Billion power network: percentage points with the wheat bag, red children, such as a number of electrical business cooperation, the current effect of how to do? Sales, transformation and other specific data how? Why are they willing to work with percentage points and not do it by themselves?

Zhang Shaofeng: At present percentage point cooperation 200 dozen electric business enterprise, the effect is not bad. Recommended engines for the wheat bag, red children and other companies to contribute more than 15% of the order accounted for. They work with percentage points, or based on the starting point of maximizing ROI, which is how to maximize ROI, whether you're doing it yourself or with someone. In cooperation with the percentage point, the ROI is greatest.

Price war is like seven punches.

Billion power network: Recently a lot of electricity Shangdou joined the price war, two think the price war will bring to the electric business industry what kind of harm? How does the electric business enterprise evade this kind of risk?

Chen: Personally think that the impact of price war on the electric business industry is more harm than good. Price war is difficult to accumulate popularity, even in a short time to attract a large number of users, is also unstable, viscous low. And the low price of marketing, only to stimulate the consumer short-term shopping mood, and no real and consumers to establish a stable relationship.

Even if the price war of electronic commerce is unavoidable, it cannot subvert the absolute value of the commodity. Otherwise, the industry will inevitably bring harm to the whole. Once low prices become a means, will undoubtedly have a huge impact on the entire industry. Small and medium-sized enterprises can not afford to bear the impact of costs, the electric business ecosystem will become a mess. E-commerce can become an important means of expanding domestic sales nowadays. With the use of electricity, we can pick up a part of our own brand. If the enterprise does not have the restraint to carry out the price war, the result is disastrous.

Personally feel that in the face of price war, electric business enterprises, especially small and medium-sized enterprises coping strategies are three.

The first is to avoid its sharpness, in the electric dealer price war during the avoidance of direct competition, can avoid promotional period, or staggered promotional items. This can avoid head-on straight hit caused by the badly beaten.

Second, the Electric Business Alliance. The joint purchase pricing, can let the electricity merchant to have the bargaining power more, in order to reduce the price to deal with prices war;

Third, product outsourcing. This is an effective way to reduce operating costs. Choose to have the strength of the experience of the electrical business outsourcing enterprises, is a wise choice of small and medium power companies.

We can be sure that, in spite of the current domestic market has several major electric power companies outstanding performance, market share lead, but for the entire Chinese electric business industry, everything is just beginning. The opportunity and risk of the electric dealer coexist, see How to choose the strategy that suits oneself.

Zhang Shaofeng: The current practice of "price-driven" rather than "value-driven" is very sad, which is tantamount to the seven-injury boxing in Jin Yong's novel, "on the Day of the Dragon Slaughter": First hurt himself, then hurt others. And once consumers are created to sell the site is a bargain, it is difficult to reverse. But for the current stage has been caught in the price war, it is difficult to immediately get out of the price war. In the last resort to continue the price war at the same time, the electric business enterprises must begin to pay attention to practice the internal strength, improve the competitiveness of outside the price, such as the goods themselves, such as the partition, service, and pay attention to fine operation, improve gross profit margin.