This paper is based on an interview with Chen Chen, a guest with 17 years experience in data analysis industry.

Interviewer: Chen Chen.

Resume: He is currently the senior director of data analysis department of Dentsu Hinkis-Merck China (Shanghai/Nanjing) and concurrently the general manager of Merck Nanjing Company. He has more than 17 years of experience in data marketing, risk analysis, quantitative models and customer relationship management strategies in the consulting industry and leading companies in the United States, Canada and China, and has a solid foundation in marketing, quantitative methods, econometrics and statistics, as well as rich project background in establishing banking/finance/retail marketing models and credit scoring models.

Q: If an enterprise wants to tap the value of data, but for various reasons, the dimensions of the data itself are incomplete, or there are many missing data. Can you share with us how such enterprises can effectively use data with many years of project experience?

Chen Chen: For advertising marketing, the more useful data dimensions, the better. Even if the enterprise's own CRM is very complete, and supplemented by other data sources, it is very beneficial to understand the existing customer groups and how to prepare for the next marketing activities.

In fact, no data provider can meet all the marketing information needs of the brand. What the brand needs is to buy and combine the most relevant high-quality data content according to local conditions. Merkle can use its purchasing power and strong partner network to help brands find the needed data on a global scale, and combine data integration and landing effect analysis to create strategic advantages for customers.

According to Merkle's common user life cycle, we divide the user life cycle into contact with potential users (customer acquisition stage), old customer maintenance (interaction stage) and retention analysis (repurchase promotion stage). In the customer acquisition stage, Merkle can combine other data sources to enrich the data dimension. For example, when we serve a well-known online English education brand, because the customer's own data is not enough to support modeling, we use the data of operators and a well-known technology company to enhance the customer's data, and the accuracy of user portrait and modeling has improved a lot.

Moreover, there are advantages in data docking with operators. Operators naturally have access to consumers. Therefore, in the second stage of this project, we will use the model to select the consumers who are most likely to be transformed, and do marketing activities at appropriate contacts by sending short messages and pop-ups.

If it is in the interactive stage, the more data dimensions, the more users can be grouped according to their behavior/state, and personalized interaction can be realized. For example, we provide the LoyaltyPlus platform solution for the NBA to help customers create an "NBA fan circle". Fan Circle is a customer loyalty system established by Merkle for NBA China by collecting, cleaning and integrating customer data from multiple data sources, so that fans can be grouped according to their behaviors and states, and they can interact with fans effectively.

At present, the loyalty system has more than 600,000 registered fans and 64% active users. The collected member interaction data will be used for customer grouping and customized services to achieve the effect of closed-loop marketing. Retention analysis, do more competitive product analysis, and understand the reasons for the loss of users.

There are different corresponding strategies in different periods of users' life cycle.

Q: Under what circumstances will the company consider using data analysis/model for optimization, how to operate it specifically, and how to evaluate the landing effect?

Chen Chen: Fundamentally speaking, data analysis/modeling/statistics are some methods used to measure data assets more scientifically. In my opinion, as long as you have data and spare capacity, you can try to perceive users from the data and improve the marketing effect.

We divide the company into two categories, one is an enterprise that pays attention to user growth, and the other is an enterprise that pays attention to customer maintenance. Of course, this division is not very strict, and many enterprises pay equal attention to both. Paying attention to user growth is to acquire customers. No enterprise needs to acquire customers, but with the rise of Internet and data, the specific methods of acquiring customers have changed greatly.

When the Internet is just emerging, people will find it simple and cheap to get customers online, such as investing in paid advertisements on search engines or doing SEO, with remarkable results.

What about now? Internet traffic is getting more and more expensive. For some specific industries, such as automobile or education, the cost of a sales lead reaches tens or even hundreds of RMB, so how to find an effective and cheap way to reach more potential consumers in the current environment is very important for enterprises.

What we are doing now is to help customers tailor accurate customer acquisition strategies and complete CRM (Customer Relationship Marketing) solutions.

First of all, understand the existing customer acquisition process and propose solutions based on the characteristics of the industry and customers themselves. After landing, you can also compare the results with historical data and what you have learned in this process, and then adjust the specific operation or implementation steps to form a closed-loop optimization result. Take a well-known computer brand customer as an example. The original way to get customers is to do online and offline activities and buy some user resources online, but we can help him do it specifically. We can select people who are interested in the brand through the model, and then do activities on this basis, saving money and effort.

Later, in the transformation stage, the original customer's method was to contact through the telephone center or in the form of direct sales. We can enrich these means, for example, we can verify the real intention and specific needs of users with data, and make personalized recommendations when contacting; Or you can use the model to group these users and then promote the conversion, which is a very good method.

Merkle CRM solution flow chart

For enterprises that pay attention to customer maintenance, we can help enterprises establish a system of user value and life cycle. The main goal of user value is as an investment standard, and there are many forms of investment in users: for example, providing more frequent and convenient services for high-value users; Provide customized contact strategies for different users in marketing activities.

We will consider many factors when determining the user value, such as the right to use, turnover, risk, marketing and service costs, transaction history and expected future profitability and income. Moreover, user value will change with its life cycle, which is a dynamic process. In this change, the marketing effect can be amplified and the correctness of decision-making can be increased. Take the computer industry as an example. We can combine users' historical purchase/warranty/online browsing behavior, give each user a life cycle stage, and then contact users at the right time.

Q: Can you give some advice to young people who are interested in the big data industry?

Chen Chen: I think to engage in the big data industry, we must first be able to settle down, master one or two commonly used data analysis tools, such as R and Python, and be able to program to a certain extent, so that we can have an intuitive and in-depth learning process for data understanding and analysis, and also exercise the mathematical and logical thinking ability of newcomers.

Of course, this is the basic prerequisite for entering the industry. Then you need to have a certain statistical foundation and business analysis ability, and you can quickly draw insights and application directions for business applications or other related professional fields from the results of data analysis. To put it simply, we should not only run out the model and draw the chart, but also infer and summarize the real story and significance from the model results and data visualization.

After mastering these basic abilities and skills, we should keep the mentality of continuous learning, constantly track and understand the latest trends and trends of the industry, and be able to horizontally integrate multi-industry directions. In addition, big data analysis majors often need to communicate and explain with different departments and different types of customers. Therefore, if you need to maintain sustainable competitiveness at the back end of your career, it is also an essential skill to communicate effectively with partners with different background levels in professional and non-professional languages.

The interview is over. Thanks to Chen Chen for sharing some project experiences on data integration/enhancement, data analysis and modeling.