Page 280 - Introduction to Electronic Commerce and Social Commerce

266 9 Marketing and Advertising in E-Commerce

Behavioral Marketing and Collaborative is greater than 35, and the customer’s income is above
Filtering $100,000, show the Jeep Cherokee ad; otherwise, show the
Mazda Protégé ad.”
A major goal of marketing is to enhance customer value
through delivering the right product or service to the cus- Content-Based Filtering
tomer. One of the most popular ways of matching ads with This technique allows vendors to identify customer prefer-
customers is behavioral marketing, which is identifying cus- ences by the attributes of the product(s) they buy or intend to
tomer behavior on the Web and designing a marketing plan buy. Knowing the customers’ preferences, the vendor will
accordingly. recommend products with similar attributes to the user. For
instance, the system may recommend a text-mining book to
Behavioral Targeting customers who have shown interest in data mining, or rec-
ommend more action movies after a consumer has rented
Behavioral targeting uses consumer browsing behavior one in this category.
information, and other information about consumers, to
design personalized ads that may influence consumers better Activity-Based Filtering
than mass advertising does. It also assumes that users with Filtering rules can also be built by logging the user’s activi-
similar profiles and past shopping behavior may have similar ties on the Web. For example, a vendor may want to find
product preferences. Google tests its “interest-based adver- potential customers who visit bookstores more than three
tising” to make ads more relevant and useful. Representative times a month. This can be done by analyzing the website’s
vendors of behavioral targeting tools are predictad.com, visiting level and activities. For a comprehensive discussion
criteo.com, and conversantmedia.com. A major method of and more information about data collection, targeted adver-
behavioral targeting is collaborative filtering. tising, and 104 companies that catch data, and so forth
(including an infographic), see Madrigal (2012).
Collaborative Filtering
Legal and Ethical Issues in Collaborative Filtering
When new customers come to a business, it would be useful
if a company could predict what products or services are of A major issue in using collaborative filtering for personaliza-
interest to them without asking or viewing their previous tion is the collection of information from users without their
records. A method that attempts to do just that is collabora- consent or knowledge. Such a practice is illegal in many
tive filtering. Using proprietary formulas, collaborative fil- countries (e.g., the USA) because of the violation of privacy
tering automatically connects the preferences and activities laws. Permission-based practices solve this problem. In fact,
of many customers that have similar characteristics to pre- empirical research indicates that permission-based practices
dict preferences of new customers and recommend products are able to generate better positive attitude in mobile adver-
to them. For a free tutorial of 119 slides about collaborative tising (Karthikeyan and Balamurugan 2012).
filtering from Carnegie Mellon University, see Cohen (n.d.).
Many commercial systems are based on collaborative Social Psychology and Morphing in Behavioral
filtering. Marketing
Amazon’s “Customers who bought this item also
bought…” is a typical statement generated by collaborative Cognitive styles that define how people process information
filtering, which intends to persuade a consumer to purchase have become a subject of research in Internet marketing and
the recommended items by pointing to preferences of similar advertising. The underlying rationale is that people with dif-
consumers. ferent cognitive styles have different preferences in website
design and marketing messages. Specifically, an attempt is
Other Methods made to connect the Web with users in their preferred cogni-
tive style. This can make one-to-one advertising messages
In addition to collaborative filtering, a few other methods for more effective. MIT designed an empathetic Web that is uti-
identifying users’ profiles are described below. lized to figure out how a user processes information and then
responds to each visitor’s cognitive style.
Rule-Based Filtering
A company queries consumers about their preferences via
multiple choice questions and uses the collected information SECTION 9.2 REVIEW QUESTIONS
to build patterns for predicting customers’ needs. From this
information, the collaborative filtering system derives behav- 1. Define and describe the benefits and costs of personaliza-
ioral and demographic rules such as, “If the customer’s age tion.