Application of data mining in CRM
From: China Computer Author: Jianghua Date: May 23, 2002 Views: 1758
There is a classic 2/8 principle in customer relationship management (CRM) theory, namely 80% of profits come from 20% of customers. So, what is the characteristics of these 20% of customers? The survey found that most companies have 20% to 50% of customers a year. On the one hand, the company is taking new customers, but the other is constantly lost old customers. Is there any way to find out, which type of customer who is lost, which type of customer gets? In the fierce competition, resources have become the key to determining the success or failure of the company. In terms of customer relationships, companies always want to establish the most stable relationship with customers, and most efficiently convert this relationship into profits, that is, customers who retain old customers, develop new customers and lock profit margins, this is CRM. Problems to focus on research. In order to achieve this goal, companies need to understand the behavior of customers as much as possible, but this understanding cannot be obtained directly by contact with customer contact, because companies are impossible to talk to customers, and their information is often unable to provide. If the company can do, it is to collect information as much as possible, with a variety of analytical methods, through disorderly, the surface-level information dug out the inner knowledge and law, which is studied by very popular data excavation technology. After digging a large amount of information, the company can make a speech of unprecedented results based on these regularities or to design mathematical models, and provide a basis for the unprecedented operation of the company. Data mining, also known as Knowledge Discovery IN Database, KDD, refers to information or mode that extracts potential application value from a large database or data warehouse. Adapting to the eagerness of the company, when the CRM scenery is infinite, the data mining has also walked out of the study room, and it has settled in business. A recent Gartner report lists five key technologies, KDDs and artificial intelligences, which will have important impact on the industry in the next 3 to 5 years. What is data excavation? In CRM, data mining is a knowledge and rule that is implicit, previously unknown, and has a potential value for corporate decisions from a large number of data about customers. Customer Features: The first step in data mining is to dig out the character of the customer. Enterprises will never be satisfied with customer information, they will not only find ways to understand basic information such as customers' address, age, gender, income, occupation, education, etc., the collection of marriage, spouse, family conditions, disease, hobby, etc. also spare no effort . Due to this reason, when talking about CRM, personal privacy has become a sensitive topic. "Gold Customer": Analysis of customer behavior analysis, the highest consumption, the most stable customer base, is determined as "gold customers". Determine the corresponding marketing investment for different customer grades. For "gold customers", you often need to develop a personalized marketing strategy to keep high profit customers. Therefore, don't expect to continue everyone equality in the CRM era. Of course, successful CRM will not make customers feel discriminated against. If you are unfortunate to find that you have received less than others, it is very likely that others are "gold", and you are "silver" or "black iron". Customer payment point: By contacting customers, collecting a large number of customer consumer information, through analysis, dealing with customers' most concerned, providing marketing activities, spending money on "point". The same advertising content, according to customer different behavior habits, some people will receive a call, some people may receive a letter; the same company will send them different information, and this information is often a customer sense. Aspects of interest.
Don't be surprised that the company will send you the most important, most satisfying, you and other customers who are similar to your similar, in the business data warehouse, I can't live "torture", I have been collectively confession. . Customer loyalty: derived by customers' persistence, firmness and stability analysis. For Gao Zhongheng customers, pay attention to maintaining its good impression, for low loyalty customers, or not to waste money, or we will cultivate them into loyal customers. Last year, a US company changed to all customers to send Christmas cards to all customers, only giving them greetings to the most loyal customers. Probably they think that since it is the customer who wants to go, why bother to waste a greeting card and a postal fee? How does data mining dig? In CRM, an essential element is integrated, forming a large, structured data warehouse (Data Wearhouse), which is integrated, structured, which is a data mining. On this basis, it is necessary to use a large amount of knowledge and methods to disclose the potential correlation and regularity to reveal the surface, dissemination, and to reveal the potential correlation and regulatory. The lateral association is the knowledge between the excavation surface seemingly independent events, such as "90% of customers who purchase product A while purchasing product A in one purchase activity). For example, the story of the classic "diaper and beer" is to use this method, find that there is a high correlation coefficient between the two, causing attention, and then finding in-depth analysis to identify the inner reasons. The side focus on this analysis of this analysis is the knowledge of the pre-rear sequence relationship of the analysis, and found that "After purchasing A products, the customer will then purchase the product B, and then buy the product C", form a customer behavior "a → B → C "mode. It can be seen that a customer is likely to purchase a printer, scanner and other accessories after bought a computer. However, if you find the model of "Shaving Knife → Pumped House → Diamond Ring" by data mining, it is estimated that the enterprise customer service department is busy with the potential link. Classification classification analysis is to analyze the data in the sample customer database, make an accurate description or establish an analysis model or to minimize the classification rules for each category, and then use this classification rule to classify other customers. For example, credit card companies divide the cardholders into different grades based on the customer's credit record, and assume the level marker with each record in the database. For each level, find them in common, such as: "The annual income of more than 100,000 yuan, the age of foreign enterprises between 40-50," overall credit records are the highest. With such a mining result, the customer service department knows the potential value of a new customer, and the customer service is put into the heart. Cluster This is a retrograde method for classification. Clustering does not have a recorded record, in order not to be partially divided into several categories, rationally divide several classes, and determine the category of each record. Its classification is determined by statistical cluster analysis method. For example, in the face of "consumption", "purchase frequency", "income level" and other evaluation indicators, there is no way to classify according to an indicator, you can use the clustering according to the natural connection between the data. Poly "becomes a few" stack "and then analyze each pile. Data mining combines database, artificial intelligence, machine learning, statistics, etc. The theoretical, standardization, standardization, standardization and analysis of data, and analyzes the "gold" required. In terms of technology, the customer relationship management system uses an embedded data mining system to automatically generate some information required. Depth data mining, also requires companies to have professional talents in statistics, decision-making science, computer science, and formulate corresponding mining rules to play the advantages of excavation systems.
