How to take customer as center for data Mining and analysis

Data mining and analysis can be said to be the fastest-growing technology in the field of information, many different fields of experts have gained the space for development, making data mining become a hot topic of discussion in the business community.

With the development of information technology, people collect data more and more rich, the accumulation of data is growing, the amount of data to GB or even terabytes, and high-level data has become the mainstream, so data mining this fusion of various analytical means, from a large number of data found useful knowledge of the method came into being, Its emergence for business decision-making provides valuable knowledge, so that enterprises gain profits, in the customer's internal demand management, data mining is playing a guiding role.

First, customer-centric data analysis framework Thinking

In the information age, three changes have taken place today, from the hardware age of the 80 's to the software era of the 90, to the customer-centric information age, which began in 2000.

We also saw in this stage of the service industry competition has also undergone tremendous changes-from the service content to win, to the service channel, to the current customer experience to win.

1. Build a business framework from a customer perspective

From a macro point of view, the business strategy is gradually in-depth customer-centric thinking, from a microscopic perspective, in the customer-centric thinking driven, the enterprise needs a set of mechanism process changes, including customer-centric data mining and analysis, customer-centric business planning, customer-centric marketing planning, customer-centric design , customer-centric performance system construction, etc. (see the Ma Haixiang blog, "Data analysis process and analysis method" of the relevant introduction).

Traditional business planning we usually only consider the logical relationship between the business, a small consideration of the customer operation from the perspective of the formation of a series of psychological reflection, behavioral characteristics, value orientation and so on, and in the customer-centric business planning, the customer needs to be detailed analysis of each operation, documented, through behavioral trajectory comprehensive analysis of customer psychological characteristics, As a result, customers can be more effective in their business, and the following two comparisons are made between traditional business planning and customer-centric business planning:

(1), traditional business planning

①, brain storms build business processes from a business planner perspective rather than a customer perspective.

②, we do not know what the customer needs, and worse, the customer may not know.

③, each link has customer churn, and we do not know what happened.

④, the customer every click is a marketing opportunity, but we missed every opportunity.

⑤, customers at every step of the missed the original he may have purchased goods.

⑥, when the customer left, we permanently lost the customer, did not leave valuable information.

(2) Building a business framework from a customer perspective

①, customer-centric thinking to build a business framework.

②, system needs to meet the individual needs of different types of customers, the core of which is data mining and application.

③ and systems need to help customers achieve customer expectations and help them identify and meet potential needs.

④, the system needs intelligence to find the best time to help, intelligent customer assistance.

⑤, System construction need to consider the future development direction of the system, its core for customer demand mining.

2, business and marketing as the center of the Data system construction

Today's marketing is undergoing a huge transformation, customers need greater participation, more interaction with the enterprise, especially in the electronic channel Interactive marketing is becoming the protagonist.

Customer requirements for enterprises are more and more high, the corresponding behavior is also changing, and the traditional push mode as the main marketing method is not only inefficient, but also make customers more and more disgusted, this fully reflected in the marketing conversion rate is low, customer rejection rate, satisfaction decline and other data.

If we look at today's advertising prices, you can find today's marketing behavior is no longer the advertising to the "new media" so simple, today's marketers must find a way out, efforts in a variety of highly interactive marketing channels to carry out high-level personalization and related communication, to create a good customer experience, and strive to improve marketing efficiency.

Take the purchase as an example, we through the customer research found several behavioral characteristics, consumers first interested in the product, at this time, although the sales staff can help customers, but not directly dialogue, the best way through the system of intelligent interaction with consumers, to help consumers make decisions, and when consumers have a real impulse to buy, Consumers are more willing to proactively communicate with customer service via IM to form buying behavior (see the Ma Haixiang Blog, "Strategies and requirements analysis for collecting customer Relationship Management data").

As a result, marketers must anticipate the entire marketing opportunity in advance, which requires us to be able to help marketers to count customer changes as they build the data system, and to meet the needs of marketers to personalize marketing tailored to the changing needs of their customers every moment of the day, helping marketers develop accurate marketing campaigns.

①, data frame construction must focus on business and marketing

②, data framework building needs to meet existing business needs and needs to meet future business development needs as much as possible

③, data framework construction focuses on the realization of intelligent interaction

④, data needs to be able to be used for analysis, judgment, decision making, use

⑤, data frame construction need to be able to reflect the changing trend of data, assist business analysis and judgment

Ii. Main events of the data analysis framework

The main Event event description classification, according to the needs of the business to carry out the necessary classification, such as the classification of customer ratings, AA grade or AAA grade estimate according to business data needs to define the need to estimate the data and data interval values, to complement and assist the business.

For example, according to customer savings and investment behavior estimation of customer investment style forecast according to the trend of data trends forecast data development direction; For example, according to historical investment data to help customers predict the investment market and other data groups according to business needs to group data, such as the purchase of a class of customers usually also buy After buying a customer, there is a B-cycle that generates a logical relationship of C-behavioral clustering data, such as having both a-and B-feature data, which can be inferred to have a C feature describing descriptive data to help extract key elements for data induction, for example, to approximate business marketing from data keywords, Memos such as complex data mining such as video,audio, graphic images, and so on.

1. Classification (classification)

In business building, the most important classification is generally the classification of customer data, mainly for precision marketing.

Usually the biggest problem of categorical data is the planning of classification interval, such as the granularity of classification interval and the interval boundary of classification interval, etc., the planning of the classification interval needs to be set according to the business flow, and the design of the business flow must take the customer's need as the core, so Ma Haixiang think the core idea of classification is to be able to fulfill the business

Because the market demand is changing, the classification is often changed, such as the savings range of the VIP customers in the banking business.

2. Estimation (estimation)

Usually data estimation is the basis of interactive marketing, based on the customer behavior of data estimation as the basis of interactive marketing has been proven to have a high business conversion rate, the banking industry usually through customer data to estimate the customer's preference for financial products, Telecommunications and Internet services typically estimate the customer's needs through customer data or estimate the customer's lifecycle.

Ma Haixiang that data estimation must be based on data segmentation and data logic correlation, the data estimation needs to have high data mining and data analysis level.

3. Forecast (prediction)

Future forecasts based on trends in data are often a powerful way to promote a product, such as the securities industry usually recommends stocks that are in good shape, and banks will assist clients in their investment to achieve a certain future expectation based on their capital situation, and the telecommunications industry usually judges business expansion and contraction as well as marketing through the growth in service usage.

Data prediction is often a common result of multiple variables, and each group of variables will typically have a number of interrelated values, and we can usually calculate the data predictions based on the relationship of each variable, and follow up as the basis for business decisions.

4. Data grouping (Affinity Grouping)

Data grouping is the foundation of precision marketing, and when data is grouped with customer characteristics as the primary dimension, it can often be used to estimate the basis of the next behavior, such as marketing supporting services and tools through the needs of the service characteristics that customers use, and customers who purchase a Class A generally have B behavior and so on (see Ma Haixiang Blog " How to analyze the user's relationship with the product through data analysis).

In Ma Haixiang's view, the difficulty of data grouping is the rationality of the grouping dimension, usually its accuracy depends on whether the grouping logic is consistent with the customer's behavior characteristics.

5, Clustering (clustering)

Data clustering is one of the key items of data analysis, for example, in the health management system through the combination of symptoms can roughly estimate the patient's disease, in the telecommunications Industry product innovation customer use of business portfolio is usually the basis of the service package, in the banking product innovation, customer investment behavior aggregation is also an important basis for financial product innovation.

Ma Haixiang reminds us a little: The key point of data clustering is the correctness of clustering dimension selection, and it is necessary to practice to verify its feasibility.

6. Description (Description)

The greatest utility of descriptive data is that events can be summed up in detail, and often a lot of subtle opportunity discoveries and inspirations come from descriptive customer recommendations, while customers prefer a descriptive approach to search, which requires technical assistance with better data correlation methods.

The difficulty of descriptive data is the extraction and classification of data elements under the large data volume, the core of which is the extraction rules and the classification methods, and the extraction and classification of features is the basis of which they can be used.

7. Complex Data Mining

Complex data mining such as Video,audio, its elements are still difficult to extract by technical means, but also from context and context to extract some elements to help clustering, such as important customers marked the high importance of video general priority should be higher.

The mining of complex data is usually solved by the standardization of data entry, the core lies in the planning of data Entry standard system, and the Ma Haixiang suggest that the initial planning is to be considered as perfect as possible, not only for the present, but also for the future.

Third, from customer demand to business

For the characteristics and needs of different customer groups, we should also have targeted data mining and analysis, with personalized services to win the majority of customers.

1, customer-centric business planning ideas

Customer-centric business planning has roughly three links: from customer research to demand mining, from demand information to data-based requirements management, from requirement documentation to business planning and design.

Customer-centric business planning not only needs to consider whether the business needs can meet the needs of the problem, but also take into account the changing business trends, business marketing focus.

2. Data mining Technology

For the mining of customer data, we can obtain the following methods:

(1), clickstream data Click Stream

①, direct Access quantity

②, visitor sources

③, visitor location

④, click Stream Tracking

(2), Outcomes data result type

①, visitors (number of initial visits, total number of visits, average return visits, concerns)

②, Page view (average number of views, total PV, visit more than one page of visitors)

③, Time (global, per capita)

④, key behaviors (e.g. registration, purchase)

⑤, conversion rate

(3) Research data

①, customer research

②, heuristic assessment, customer experience testing

③, customer attributes (database analysis)

④, customer expectation analysis (from data to service)

(4), competitive data competitive

①, "Polygon" Data Measurement (VW analysis)

②, Network Service data measurement (industry analyst)

③, search engine measurement (public opinion analysis)

3. Data analysis Technology

For data analysis technology, we can be divided into primary data analysis and advanced data analysis of 2 kinds:

(1), Primary data analysis

①, click Density Analysis Click Density

②, Visitor Primary Purpose Visitor Primary Purpose

③, Task Completion Rates job completion rate

④, segmented Visitor Trends customer tiering

⑤, multichannel Impact analysis channel

(2), Advanced data analysis

①, Customer Value Group properties

②, customer feature Group properties

③, data estimate combination

④, data expected value combination

Combination analysis of ⑤ and clustering

⑥, customer deep-seated research

4. The idea of data interaction across channels

①, cross-channel data interaction for General Service or marketing purposes.

②, cross-channel data interaction must be customer-centric.

③, cross-channel data interaction can give customers a three-dimensional experience, effectively enhance the brand experience.

5. Data-based interactive business planning

①, data interaction based business planning objects are generally a series of products or services chain, usually widely used in communications industry, banking, insurance, retail and so on.

②, business planning based on data interaction must be customer-centric, analyze the timing of customer demand, and intelligently match products or services, and its implementation takes data mining as the core.

6. Interactive marketing planning based on data

Interactive marketing based on data mainly refers to the interactive marketing, the core idea is to analyze the customer's specific timing needs, and according to the need to recommend related products or services to meet customer needs, widely used in various industries.

Business planning based on data interaction also needs to be customer-centric, analyzing the timing of customer demand, and intelligently matching products or services, with the same data mining as the core.

7. Data forecast

Data analysis: Compared with the use records of A and B services, customers who use a service will receive less than B service at 1 months, while 3 months of service can generate more revenue than B services.

Interactive Marketing: It is recommended that customers who need 1 months of service use B service, it is recommended to use a service for 3 months.

Business Innovation: Development of non-a non-B type C services for customers requiring two months of service.

How to use customer-centric data mining and analysis (GO)