Download Database Marketing and Method of Customer Behavior Analysis

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Database Marketing and Method of Customer Behavior Analysis
WU Qinghui
University of Science and Technology Beijing P.R.China,100083
[email protected]
,
Abstract: this essay mainly discusses the database marketing effect and character of financial service
industries; researches database marketing method and application of financial service industries.
Key Word: financial service industries; database marketing; method and application;
1 Introduction
On the international market, product marketing and managing mode has developed since it came
about, and has produced new characters, that is marketing aim layering, marketing decision integration,
marketing object diversification, marketing method dimension, marketing channels integration,
marketing exchange bidirection, marketing process informatization, marketing service systematization,
marketing activity culturalizing and marketing plot novelty etc. On the above base, the new marketing
strategy emerges, for example, the megamarketing strategy, relationship marketing strategy, corporate
marketing strategy, network marketing strategy etc. proposed by Philip Kotler based on the traditional
marketing.
Because of customer and product specialties in product marketing management on behalf of
financial industries, during the optimization of marketing management process, there emerge some
difficulties as follows: the difficulty in prediction of benefit, the complexity of marketing plan, the high
quality in customer segmentation and product positioning, the hardship of prediction on marketing
results etc.
According to the above specialties, many enterprises conduct active experiments in multiform
marketing management modes as follows:
(1) Cross marketing which is a marketing mode based on bank product relative analysis.
(2) Affiliate marketing is a marketing mode to expand product scope through win-win cooperation,
strategic union and external resources utilization.
Nowadays, it has come into being that database marketing mode has been introduced to the
enterprises with leading management ideas. Database marketing comes from the concepts of
relationship marketing, builds on database and data mining techniques, and it is a new and advanced
marketing management mode no matter on the side of marketing ideas or on customer management.
2 Optimizing marketing management process—Database marketing
Database marketing is the way to conduct focused and efficient customer marketing, based on the
analysis of historical data and information in the aspect of products, market and customers, utilizing the
strong and special capabilities in data organizing and analyzing. Thus database marketing has played a
significant role in financial industry that banks services produce enormous data and deal with large
numbers of customers.
2.1 Function and feature of database marketing’s
As a emerging and precise marketing mode, database marketing, through precise positioning of
target market, analyses and designs precisely on customers and products, and then anchors segmentation
group. The key is to control enormous database sources, to conduct marketing on target customers, and
to coordinate with the efficient implementation and channels. Compared with other traditional marketing
modes, database marketing is more accurate, interactive, controllable and consecutive. (refer to the form
below):
321
Table1
comparative analysis form of bank database marketing mode
Element
Database Marketing Mode
Traditional Marketing Mode
Target customer analysis base Personal target and information
Group target and general information
Customer Group
Valuable customer
All customer
Information communication
Two-way interactive
One-way
Sale channel
diversified
Single
Marketing way
Focus on target customers
Universality on general customers
Service way
Personalize
Popular
Control level
Entire control
Not control
Maneuverability
Limited without time & place
Limited by time & place
The function of database marketing is mainly listed as follows:
(1) Precise customer segmentation and product positioning.
Utilizing data organizing and analysis techniques, we can conduct exact prediction and realize
precise positioning due to database can make commercial bank focus on the customers who are more
efficient. Nowadays, 56 % of American enterprises are building database, and 85% of American
enterprises consider that they need database marketing to strengthen their competitiveness.
(2) Providing service differentiation and steadying customer group
Now, product’s service and service’s product have been highly integrated; we need to put
customer’s value notion into the enterprise whole business, and work on customer-orientation. Find,
satisfy and continuously exploit customer’s need is the basic way to make benefit for commercial bank,
build and use customer database, so that we can master customer’s need timely, which provides precise
information for producing new product, makes commercial banks design and produce product according
to customer’s need, start direct service and enhance customer’s loyalty to company and product. From
the practice, it shows that the success ratio to develop new customer is 30 percent, but the cost is 4-5
times than developing the existing customer. If enterprises invest 5% resources to maintain and exploit
existing customer, they will improve about 70% benefits.
(3) Reducing cost and improving marketing efficiency
Today, customers are increased pursuing characterization and personalization. Because of the
indivisibility of traditional marketing, it has caused waste a lot of marketing resources, the bank need a
more efficient and new marketing management mode, in order to make the product satisfy target
customer’s need, the bank need to segment market according to this character and change. For example,
direct mail without selecting customers through database techniques, the rate of feedback is only 2% to
4%. By contrast, feedback rate for application on database techniques surf to 20% to 30%.
(4) Favoring to develop discriminate competition
To employ database marketing mode, commercial bank and customer can build close relation with
secret inter-bank competition that would avoid rivals’ attention. Meanwhile, mail database can provide
sufficient two-way contact between banks and their customers and maintain customers’ emotion nexus
to strengthen the banks competitiveness.
To discuss and research on database techniques under the new situation plays a significant role in
promoting enterprises development. Marketing mode based on database set aim on satisfying customer’s
need. On method, it makes marketing object more direction; on mode of thinking, it makes marketer
change basically to enrich marketing method and design more efficiency marketing plot. Taking bank
card for example, the banks that research and utilizing on database marketing in advance are holding
leads in financial product marketing (refer to the form below).
Table 2 comparative form of quantity of bank stuffs engaged in database marketing and bank’s business
volume
Bank
Name
Bank 1
Bank 2
Bank 3
Bank 4
Bank 5
Quantity of staff
100
1200
2000
800
400
2005
Annual Volume of Card
136,800
1,641,600
2,736,000
1,094,400
547,200
322
Quantity of Staff
1000
2000
3000
1500
1000
2006
Annual Volume of Card
1,368,000
2,736,000
4,101,000
2,052,000
1,368,000
Table 2 displays the differences among the above banks between quantity of staff engaged in database
marketing and annual volume of their issued bank cards. Bank 3 which develops database marketing
research earlier holds a lead in business volume.
300000
250000
266330
200000
150000
103670
100000
35750
50000
0
bank3
bank2
bank1
17707
6510
bank4
bank5
Figure 1 Comparative Graph on Banks’ Database Marketing Results
The above graph shows transaction volumes compared among the above banks. Bank 3 does well
in database marketing research and utilization, its transaction volume is several times as much as other
banks. It displays that database marketing has great promotion not only on increasing product quantity,
but also market excellent customers and increase using rate and amount of business transactions.
Therefore, database marketing’s research and popularization has a significant role in business
development, customer analysis and service etc. it should be valued and supported by the decision
maker.
3 Establishment and application of customer behavior analysis system
Enterprises in the course of selecting and setting up marketing strategy must analyze and segment
customer group deeply. Taking financial industry for example, the industry always has thousands of
customer information, but most of information are distributed in various business systems. In general
information, it mainly records customer’s status, and records less information about investment
preference, financing habits and culture level etc. Thus it is difficult to analyze customers in this kind of
enterprises, they should introduce the data mining theory and technique and solve this problem
scientifically. They should segment customer groups, carry out the relevant marketing plot, select
marketing channels and improve market operation.
During analyzing customer behavior and designing marketing strategy, enterprises should follow a
few steps:
Step 1. Definite the goal of marketing plot, which is to search customers fit for marketing standard
among various customers’ information, and then work out different marketing plots according to
different customers.
Step 2. Search the relevant and important data property set in relevant database.
Step 3. Analyze and gain the most important data property
Step 4. Take this data property as model variable, and calculate this model.
Step 5. Differentiate customer group, design marketing activity program
Step 6. Check model’s veracity, examine program’s efficiency
Step 7. Summarize and analyze activity’s experience and lesion.
3.1 Pre-preparation for data
In analyzing customers, at first, make sure data’s property. Because large enterprises’ database
species and quantity, different customers and transfer information are in different database, enterprises
323
should refer to experts judgment, salesmen judgment and classification analysis etc. from some relevant
or not relevant database to select the directly data property, and then take this as foundation to work out
the analysis of customer or transaction. For example, financial industries’ database should include below
information, customer information, account information, card materials information, transaction
information and score information etc. In financial business marketing, it is the simplest situation to
differentiate different kind of customers, research the characters such as age, education, sum of
transaction etc. and formulate marketing plots separately. Thus we can take customer information
database and transaction information database as basic database, among which, customer database
includes the following attributes:
Name
Age
Gender
Education
Table 3
Attributes of transaction database
Family Add. Post Add.
Credit line
postcode
transaction
Expiry Date
Card number
Transaction
Date
amount
company
Customer
code
position
Deposit
date
3.2 Techniques and methods of system establishment
3.2.1 Definite analyzing goal
In numerous customer and transaction information, analyze customer group specialty who obtain a
particular consuming ability, and design different marketing plots refer to its own specialty. The
common marketing activity is that analyzing different consuming customer group in numerous customer
data; take their own consuming value as the activity standard to design the directly consuming and
transaction activity standard.
It is very important to increase the efficiency and definite of model analysis through analyzing to
gain the most main data property. We can take information plus as classification measuring standard,
and choose the largest information plus’s decision property. At first, calculating expecting information
content about classification, according to formula (1):
Gain each decision property’s expecting information content. By formula (2)
Gain property A toward classification’s expecting information content. Through formula (3)
Gain property A as decision classified property’s measure value, that is information plus.
Calculating each decision property’s information plus, and take the property which has the biggest
information plus as decision property. In the process of financial product marketing, we should choose
property “transaction amount” as classification mark, and property “gender”, “education”, “age” as
decision property assemblage, which consist below basic information form of customers.
Table 4 Basic Transaction Information Form of Customers
Gender
Female
Female
Female
Male
Male
Male
Education
Middle School
Middle School
University
University
University
Middle School
Age
20-35
35-50
20-35
36-50
20-35
36-50
324
Transaction Amount (person/monthly)
5000-8000C1
1000-3000C2
3000-5000C3
1000-3000C2
3000-5000C3
3000-5000C3
Male
Middle School
20-35
5000-8000C1
According to sample Classification Mark, distribute three types (M=3), there are seven groups in sample
data assemblage S, and C1, C2, C3 type match along with r1=2;r2=2;r3=3 in Sub-assemblage
R1;R2;R3.
According to each decision property to calculate expecting information content, 1) “gender”:
When gender=female,
When gender=male,
Property “gender” value:
Gender’s information plus:
2) Education’s information plus:
3) Property “age” value:
325
In conclusion, property “age” has the largest information plus. Thus, choosing age as the main
variable.
3.2.2 Choosing variable, building mode.
After making sure the above data properties, we should choose it as mode variable to calculate the
mode. Since customer’s consuming transaction shows definite distribution rule on monthly amount
distribution, that is 50-100; 100-500; 500-1000; 1000-3000; 3000-5000; 5000-8000; 8000-10000;
10000-15000 etc. Thus take above eight distributions as object properties xi, i=1,2,…8. Because in
marketing activities, customer’s recognition and activity’s operability effect directly activity result,
through many times experiences, it certifies that it is not good to have too many consuming transaction
level, or it will emerge many disadvantages, such as difficulty to remember customer in operation step,
answering increased of customer’s consultant toward customer service department, advertisement cost
increased, pleasant feedback and purchase increased etc. Thus during in actual operation, it is better to
choose two levels, that is Number K=2.
Table 5
Property
Transaction(RMB)
Age
x1
Customer Basic Transaction Information Form
x2
x3
x4
x5
x6
x7
x8
50-100 100-500 500-1000 1000-3000 3000-5000 5000-8000 8000-10000 10000-150000
18-20 20-25 25-30
20-35
20 -35
36-50
40-50
40 -50
Since activity target is to maximums raise consuming transaction, thus it is better to choose the lowest
level as each consuming level’s value, it is suitable to use average way in age distribution, the above
shows below:
Table 6 Customer Basic Transaction Information Form
property
Transaction(RMB)
Age
x1
x2
50
19
100
23
x3
x4
x5
500
28
1000
23
3000
23
x6
5000
44
x7
x8
8000
46
10000
46
Divide eight objects into K types, firstly random choose two types, take each object as the center of
one type, take the rule of the most near to center as the standard, and distribute other objects to each type.
After finishing the first time distribution, take property’s average value in each type object as the type’s
new center to redistribute the object, repeat this distribution until there is no changes, and then gain the
final k types.
Step1. Random choose two objects, each object as a type’s center, which stand for two types be divided.
Step2. According to the rule of the most near to the center, distribute other objects into each type.
Step3. For each type, calculate all other object’s average property vale; take the vale as new center
Step4. According to the rule of the most near to the center, repeat distributes all objects to each type.
Step5. Return to Step3 until no changes.
Thus {x1; x2; x3; x4}, {x5; x6; x7; x8} become the final two types.
3.2.3 Differentiate customer group, design marketing program
According to the above calculating, we gain two types target customer group, that is monthly
transaction amount between [50; 3000], (3000; 12000], and ages [18; 35], (35; 50] customer group. In
designing marketing program, we should completely take these two types of customers’ consuming
326
ability, preference and habits into consideration, when the beginner of consuming transaction amount
designed, take the two-type customers’ transaction average value as reference, that is taking 450RMB
and 6500RMB as the activity beginner, when customer’s transaction reaches the above amount, reward
will be deserved.
After calculating the mode, examine the efficiency of activity program to certify whether this
calculation is correct or not. When make out the rewards towards different level customer group, we
should full operate the calculation and analysis of the above modes, and make the distribution of
customer group more definitely, and learn more about customer’s preference. On this basis, the reward
program will be workable. These data analysis is very important for financial industries to master
customer’s preference and habits, they are also the foundation for the enterprises to attract and
encourage customer transaction behaviors.
4 Conclusion
Before the enterprises make the marketing activity decision, they must use data analyzing
technology to research database information; analyze customer organization, customer transaction
behavior and customer preference etc. when enterprises’ relevant policies are decided, they should make
full use of technology analyzing methods such as data exploit, meanwhile, combine with their own
industries and market. Especially towards the industries which has big data and distinct differentiation of
customers, they should make use full of data exploit technology such as clustering and classification,
divide customer level and make sure transaction preference to raise marketing success rate and reduce
marketing cost.
References
,
[2] kusiak,K.H.Kernstine,J.A.Kern,et a1.Data Mining:Medical and Engineering Case
Studies[A].
[3] V.Ciesielski,G.Pals a.Using a Hybrid Neural/ExpertSystem for Database Mining in Market
Survey Data[A]. Proc. Second Intel. Conference on Knowledge Discovery and Data Mining(KDD
一96)[C].Portland:AAA1 Press,1996.38.
[1] Sen Wu Xuedong Gao. Data Base and Data Mining :Beijing Industry Press,2003. 9.
[4] Fayyea U M, et al, Knowledge Discovery and Data Mining: Towards and Unifying Framework.
In: Proc. of 2nd Intel.Conf. On Knowledge Discovery and Data Minging(KDD-96),1996, AAAI Press,
66-77.
[5] Chen W M Data Mining an Overview from An Data-base Perspective IEEE Trans on
Knowledge and Data Eng l996(8) 866
.
,
:
.
: .
327