Marketing application Data Mining who's doing better?

Data Mining background

The current era is the internet era, the deep application of the Internet has covered all walks of life, young and young. Anyone, no matter what occupation, what business model, product or service, if you want to effectively open up the market, arouse concern, wake up customers, can not leave the Internet platform and alone good. In a word, if you ignore the influence of the Internet, any business, any industry can not have big development, may not even survive. That sounds absolutely brutal, but it's basically true. As a professional in data mining marketing applications, if you are unfamiliar with the marketing application of the Internet, it is not just "regret", it is likely to become a "short board" that seriously affects your professional ability, because you are not living in the "Now" (you most live in the era before the "Internet"), It was a very "distant" history, like the Stone Age. In a word, if there is no "Internet application" in your field of expertise, you are not a modern professional.

Since the situation is so cruel, so active or passive, everyone in their respective areas of expertise, should try, familiar with their professional application in the Internet practice. Writers, who may consider publishing their work on the web; singers, who have already sold their music on the internet, have a large number of enterprises in the fast-food business in the network to set meals, ticketing companies are also vigorously develop network sales channels. As a data mining marketing application professionals, also should be "clever, digging network", so that this article, the current more mature network user behavior Mining Marketing Application Small summary. My thinking and summary of Web mining is mainly from the perspective of the bystander to learn and reference; In the years to come, with my web mining project practice gradually input, I believe that the field of thinking and summary will be more vivid, more real and more valuable. In view of this, at this time it is more necessary to the current theory of some ideas and insights into the text into the blog, leave it to a year after the real from the Web mining project to get a new understanding of the comparison, let the practice to prove this period of paper "Web Mining Marketing Application Summary" is no value of the armchair, Or is it true that "the right theory can successfully guide practice"? Oh, life is everywhere not contradictory, life everywhere not dialectical!!! Through contradictions, life is free and easy, learn to dialectical, LIFE has improved!!!

Three steps of network mining

Typically includes three chunks of content (Web content mining, web structure mining, and the most widely used web usage mining that is directly related to marketing applications), this article deals only with this web usage mining, which is the most direct and close to marketing applications. The following is the example of the business website, which shows that there are more mature ideas and series of methods and models from the perspective of marketing application.

First of all, from the website business Operation management of some characteristics of the indicators to analyze. All walks of life are suitable for the characteristics of the industry characteristics of the indicators (KPIs), through the analysis of these KPIs, tracking, you can quickly and accurately from the macro to judge the efficiency of the operation of the enterprise. There are some similarities between the consumer Web site and the traditional retail industry (both retail and profit-making for consumers), but the difference between the business website and the traditional retail industry's personalized index is the basic feature of the industry, must pay attention to, focus on analysis. These key indicators, features include: flow registration ratio, shopping cart ratio, order conversion rate, page views, order average browsing time, customer unit price, repeat purchase rate, and so on.

Next, from the site of the monthly, quarterly, annual comprehensive summary data comparison, from the macro point of view of the operation of the continuous period of time operating efficiency, customer changes, profit trends, product trends, consumption changes and so on (product, profit, customer latitude to start analysis). This macroscopic statistical summary analysis is relatively simple, but very effective, can quickly discover the development trend of the enterprise in recent years, the problem, can even lock the core value of the customer's group size and threshold indicators, such as the 2080 principle in the specific definition of the enterprise, For example, after the customer registration specific promotional stimulation to generate consumption of the time period of the clear definition, and even the general rule of customer churn and time, access to the Web site analysis, and so on.

The third step, on the basis of the above two-step simple analysis, aiming at more in-depth marketing problems and customer relationship management, we can consider the application of data mining. At present, the most common methods in data mining applications are cluster analysis, association analysis, and the application of in-depth predictive models (such as logistic regression, such as decision tree application, etc.).

Enterprise Specific Marketing Application

1. Division of consumer groups, to the website users according to different marketing requirements of the multiple Latitude index division, find out the consumption characteristics of the core consumer groups (especially the characteristics of network behavior), and accordingly take targeted marketing measures and service measures to meet; this kind of clustering analysis can be used for user Association, Interest Association, Read recommendations, product recommendations, and so on.

2. Analysis of consumption characteristics of a certain class of consumer groups, find out the related high profit contribution of the commodity portfolio, and accordingly formulate targeted promotion measures, marketing promotion, product strategy, price bundle strategy, etc., similar to the retail business inside the basket analysis;

3. Profitable consumer groups consumption characteristics analysis, loss analysis, loss characteristics analysis, life cycle analysis, cross-selling analysis, and so on, based on these analysis of the clues to develop the corresponding marketing measures, customer care (retention), potential excavation;

The above example is some of the most common web mining marketing purposes, practical applications in accordance with the actual business model and actual data resources, can launch the ever-changing expansion of applications, it is impossible to list out.

Data Mining Application

To put it another way, from the Internet industry's hot terminology, "product recommendation Engine" and "user-oriented" These two popular applications is to enhance the core value of the site is an important way, in fact, can be through the above data mining technology to successfully answer, other applications including Web site path design and optimization (mainly using link Analysis technology), fee-based product Classification marketing, and so on general Web site marketing operation and management of many major problems and areas, are able to use data mining technology to effectively solve. As for each of the above mining algorithms in the practical application of specific considerations and mature routines, now there are some clear patterns and shortcuts, for example, in cluster analysis mining, the most mature commercial applications are basically based on the network user's browsing frequency data indicators to analyze ( such as the amount of consumption, profit, stage time, such as the number of browsing, and so on; For example, in many large Web sites are more than hundreds of thousand or more pages, the use of the classification method can effectively compress the page types, so that the results can be more effective to promote the application of guidance practice.