Five applications of e-commerce data operation

Five applications of e-commerce data operation

Make websites more attractive

The Design and presentation of website pages are of great value, because for Internet enterprises, the first thing they face directly is the pages on the Internet. If the page design is unreasonable or the user experience is poor, it is impossible for the customer to retain and make any purchase.

The user experience involves a wide range of content, from product display, browsing, ordering, customer interaction, and so on, and even where the content on the webpage should appear, the color, shape, and location of the purchased button are also part of the user experience. In short, the so-called focus on user experience is to carefully consider the needs of the customer in all aspects of the website, and try to meet.

The arrangement of page content on e-commerce websites is similar to the arrangement of items on shelves in supermarkets. Placing items associated with certain support and trust together may help sales, using association rules, you can learn how to dynamically adjust the structure of the site for the customer, so that the links between the associated files accessed by the customer can be direct, making it easier for the customer to access pages of interest. If the website has such convenience, it can leave a good impression on the customer and increase the probability of next visit.

We can optimize the link structure of a web site from two aspects: first, we can discover the relevance of customer access pages through web log mining, in this way, links are added between closely connected webpages to facilitate customers' use. Second, the web site is optimized based on the page Access frequency and access track through web log mining.

Let's look at an example. Suppose we find that the access track on the page is like this (see Table 1-4 ).

From table 1-4, we can find that page Z becomes the end point of the trajectory at 80%, and after page Z is accessed, the probability of no other pages being accessed is 100%, or the exit rate is 100%. In this case, we need to evaluate and analyze page Z to see why the customer no longer accesses other pages after accessing page Z. Even in the ideal situation, page Z is a shopping cart page, we still need to consider why customers will not stay on the website to view other content. In short, we need to optimize page Z so that page Z is no longer the end point of the customer's access track.

In Chapter 7th and Chapter 8th, we mainly talk about how to introduce effective traffic to websites.

Convert potential customers into real customers

For an e-commerce website, it is very important to understand and follow the recorded customer groups, but it is also critical to discover potential customer groups from a large number of random visitors. If you find that some visitors belong to potential customers, you can implement marketing strategies for such visitors so that they can become our new customers as soon as possible. Data Mining of customer access records allows you to use classification technology to find potential customers on the network. Classify existing visitors into three types: new visitors, occasional visitors, and regular customers. For each visitor, the classification model can be used to identify some common attributes of the customer and the existing customer, so that the new customer can be correctly classified. Then, based on the classification, determine whether to treat the visitor as a potential customer. For new visitors, the information we can collect is relatively limited. Without other associated website information, there is only some log information. For occasional visitors, we can have a lot of information through the stay time and depth of the two visits.

We can also summarize valuable customer sources and analyze the source data to find out their commonalities, so as to increase investment in the sources with the most traffic direction or the most cost-effective sources. For example, through data analysis, we may find that in online Alliance (website Alliance) advertising, the most cost-effective advertising for some types of websites in a certain period of time, therefore, we can change our advertising strategy so that we can find more potential customers under a limited budget. We may also find that the proportion of natural traffic is relatively high, so we will need to pay more attention to Seo.

As an e-commerce website, the ultimate goal is to earn money. While considering the conversion from potential customers to real customers, we also need to consider the final consumption that these customers can provide for us in the future. Let's look at a foreign e-commerce website's customer value diagram (see Table 1-5 ).

From table 1-5, we can see that the conversion rate, repurchase rate and average total purchase value of traffic from different sources are quite different. Here, direct access to all aspects of data is the best. In addition, the conversion rate is the best promoted by Google adwords keywords; the repurchase rate is the best traffic from Google Adsense and Google natural search; if you look at the average total purchase value, Facebook and Google naturally search traffic data is the best. In Chapter 9th, we will talk about the application of traffic conversion.

Explore the value of old customers

Finding a customer is the first step, and it is more important to cultivate this customer into a valuable old customer. An important application of data analysis is to maximize the value of old customers.

The 80% law says that 20% of the company's business revenue is usually from of customers, and the sales promotion to new customers is several or even dozens of times longer than the sales promotion to existing customers. Through web data mining, we can find out what kind of customer groups have purchased products on the website within what time period, what is the average expenditure, and what type of their favorite products are, for new products which customers may purchase and which are the most important customers to retain, so as to conduct personalized marketing and humane care for them.

Here is a simple example.

The consumption times, average consumption amount, and total consumption amount of each of the five customers in Table 1-6 far exceed the average customer's number. Which of the following is our most important customer? Alan is our best customer from the perspective of the number of consumption and total amount, but Celine is the most valuable customer from the average consumption amount, followed by Bill.

In the next chapter, we will introduce in detail how to use data to mine the value of old customers.

Design and Application of Recommendation System

Establishing an e-commerce recommendation system is a mature e-commerce website and a general e-commerce website. It is also an important part of e-commerce operations. Amazon is the best in recommendation systems.

Simply put, a recommendation system is a system that recommends products to customers or provides information to guide customers to buy products. The recommendation system can simulate the customer's shopping guide process based on other customers or the customer's information to provide personalized services for the customer. The recommendation form includes predicting the degree to which the customer is interested in a product, recommending the product to the customer, or providing personalized product information based on the customer's interest characteristics and purchasing behavior.

The recommendation system can transform viewers into buyers. Sometimes people just look at the content of the website and don't mean to buy it. Then, the recommendation system can help customers find the renewal points of a product they are interested in and are willing to buy, to promote consumers to form purchase behaviors. The recommendation system can recommend customers to buy related products based on the products they have already purchased, or buy products with higher profits for sellers, to increase cross-selling and up-selling ). The purpose of the recommendation system is to build loyalty because customers are more willing to shop on websites that best meet their own needs. In section 3.4 of this book, we will introduce the principles of e-commerce website recommendation systems. In Chapter 10th, we will further introduce the application of recommendation systems.

Tailored products for different customers

E-commerce enterprises (platforms) can learn the personal interests of visitors, better understand the needs of customers, and refine the market according to various information, even providing personalized products for the unique needs of each customer is conducive to acquiring new customers and improving the satisfaction of old customers. In order to make the network information mining technology better applied, merchants must record all the characteristics and terms and conditions of visitors. When a visitor continuously accesses a website or its associated website, the visitor data will gradually accumulate.

Strictly speaking, providing personalized products for different customers is also an extension of the E-commerce recommendation system, but the recommendation system is the ultimate solution, therefore, it is our ultimate dream to provide a website tailored to each customer. What is different from the recommendation system is that the customer personalization we mentioned here is more about the personalized modifications made to webpages, these modifications are based on the customer's purchase behavior, browsing behavior, search behavior, and other customer data information. In the ideal situation, each customer's webpage is customized for the customer's personal circumstances.

According to the statistics of the Recommendation System Company brainsins, 77% of consumers think that personalized services are very valuable, and providing personalized products for each customer is the ultimate goal of e-commerce websites.

This article is excerpted from the book "Data Mining: an e-commerce operation breakthrough"

Tan lei

Published by Electronic Industry Publishing House