Website Databases—A Marketer’s Dream

Data base – an internal or external collection of raw data arranged logically and organized in a form that can be stored and processed by a computer.

Data base marketing – the practice of maintaining customer databases with customer data such as their names, personal contact information (phone #, mailing address, e—mail address, past customer purchase behavior history (inquiries, responses to promotional offers, and purchases), and demographic and financial information.

Databases are essential to relationship marketing – the process of creating, maintaining, and enhancing strong, value-laden relationships with customers and other stakeholders. This is based on offering superior customer value and satisfaction. This is important since in today’s highly competitive marketplace it costs many times more to acquire a new customer than it does to hang onto an existing customer.

Sources for building a customer database include 1. Customer transaction records (customer name, customer contact information, item purchased, and price paid),

2. warranty cards, and 3. secondary data.

Marketer Websites serve as a great source for capturing customer transaction records to build a database.

Two types of marketer Websites are:

(1) Web retailing. A web merchant (e-tailer) has access to data about its clients that would make its bricks-and-mortar counterparts cry. Besides names, personal contact information, and past purchase behavior, a Web merchant can record a customer’s actions as she moves through the e-tailer’s site, recording not just purchases but also “window shopping.’ This monitoring of a customer’s every move through the virtual aisles of an on-line store is done via cookies – a text file that a content provider can place on a user’s computer in order to identify the user when she revisits the web site.

Privacy advocates are concerned about consumers’ loss of their “right to privacy’ from merchants’ use of cookies. However, even they admit that there are some advantages of cookies for consumers. For instance, cookies can store passwords, sparing users the hassle of having to identify themselves every time they go to a Website, and they make on-line shopping carts work. Despite what some net users believe (including Zikmund – p. 46) a site can only read cookies which that site previously put on a user’s system, not any other cookies the user has collected around the web.

Cookies help merchants by recording where on the site a user goes and how long he lingers there. The end result is a file that allows the web merchant to determine what the customer is most likely to purchase next—and then offer inducements to make it happen. E-tailers can then target appropriate promotions to return visitors, and they can even rearrange the entire store layout in real time, placing advertisements of potential interest to the visitor in “high-traffic” locations.

(2) Web advertising. A company can track which goods and/or services its customers are most likely to purchase and when. It can then purchase keywords for these goods or services on Internet directory sites like Yahoo. When a Web surfer searches for information on the products, an advertising banner or link to the e-tailer could accompany the list of results.

Because the marketer can collect such massive quantities of data, data mining is often used to make sense of it all. Data mining is the use of massively parallel (powerful) computers using statistical and other advanced software (e.g., probability sampling, descriptive statistics programs, and multivariate statistical programs, as well as more advanced tools such as neural networks, genetic algorithms, and case-based reasoning systems) to dig through mountains of data to discover nonobvious patterns hidden in the database about an organization’s customers and products. For example, neural networks are computer programs that are a form of artificial intelligence—they mimic the way the human brain processes information so as to become capable of learning from examples to find patterns in data. As scientific empirical investigation does, it uses induction, generalizing from specific examples. E.g., American Express creates a set of “purchase propensity scores” for each cardholder, and thereby matches offers from affiliated merchants to individual cardholders' purchase histories and encloses these offers with their monthly statements. E.g., Wal-Mart develops “personality profiles’ for each of its roughly 3,000 stores so store managers can determine appropriate product mixes and merchandising strategies.

Since data mining is expensive and complex, it is usually outsourced to firms like DataMind, Oracle, and Information Builders. Customers send them the database and the companies perform the analysis.

Some applications of data mining:

(1) Market basket analysis -analysis of anonymous point-of-sale transactions to identify correlated variables. This enables retailers and direct marketers to spot product affinities and develop focused promotion strategies that work better than “one-size-fits-all” strategies. E.g., Camelot Music holdings identified a group of high-spending, 65-plus members of the store’s frequent shopper club were also buying lots of classical music, jazz, and movies. However, a large percentage were also buying rap and alternative rock. Data mining revealed that these were grandparents buying for the grandchildren. Now, Camelot tells the senior citizens what’s hot in rap and alternative rock, as well as in traditional music. E.G., the American Express selective envelope-stuffing strategy above. E.g., grocery stores mining checkout scanner data found that men who buy diapers in the early evening also often pick up a six-pack of beer. They could promote the two together (“Pick up a pair of six-packs—diapers and beer. Baby and brew – it’s all for you!”

(2) Customer discovery data mining - enabling the marketer to predict who will be categorized as, e.g., a good vs. bad credit risk.

(3) Sequence discovery – sequence patterns of behavior are detected, such as purchases of complementary products.

(4) Customer acquisition – in a two-stage process, direct marketers first discover attributes that predict customer responses to offers and communication programs (direct mail, catalogs, etc.). In the second stage, the attributes of customers that the model says are most likely to respond are matched to the corresponding attributes applied to rented lists of noncustomers.

(5) Customer retention – data mining identifies those customers who contribute to profitability but are most likely to defect to a competitor. Vulnerable customers can then be targeted with special promotional offers.

(6) Customer abandonment – some customers cost more than they contribute and should be encouraged to take their business elsewhere.