There are many ways to implement collaborative filtering. Collaborative filtering goes something like this. John likes audio books by David Sedaris. Other people who have bought audio books by David Sedaris also bought books by George Carlin. Therefore, the so-called recommender system at Amazon or at Audible books would make a recommendation to John that he should buy a book by George Carlin. Collaborative filtering systems can also include rating systems; in fact, Amazon and a number of other online retailers will try very hard to get you to help them by asking you to rate a product you have just bought. They will use the ratings to develop an entire web of recommendations to many of their customers and to retarget you with similar products. Here is another example of collaborative filtering in action: John bought and gave his new Kindle e-book reader a five star rating. He and many other buyers of the Amazon Kindle also bought a leather case. The recommender system will subsequently recommend a leather case to everyone who subsequently buys the Kindle.
Collaborative filtering can also involve price differentiation and price personalization. If the person who buys the Kindle does not buy the leather case at the same time, then the recommender system will send an email indicating that the leather case is on sale or wait until the Kindle customer logs back onto the system and then present the customer with a discounted price on the leather case.
Auctions are also a form of personalized pricing. Theoretically, an auction participant will bid up to their reservation price or their willingness-to-pay level for a product. Figure 2.2 illustrates that the revenue generated by offering a product at a single price of $30 will generate $900 in revenues. As illustrated in Figure 2.3, the use of an auction could theoretically generate revenues of $1,400. Auctions permit sellers to price discriminate according to the customers’ willingness-to-pay. Some individuals will bid $10 or $20 and others will bid $30, or $40 or more. As a result, a seller could theoretically generate additional revenues of $500 by offering multiple units of a product at an auction. Differentiation in Action will illustrate in detail how this revenue is generated using versioning.
Developing personalized pricing is an idealized goal for producers because the potential opportunities for revenue generation are exceptional. However, because it is difficult to accomplish in practice, producers often turn toward second- and third-degree price discrimination to generate additional revenues.