How often do Web site recommendations work for the average customer? Yahoo.com offers advice as customers browse its music and video catalog and suggests selections based on what other shoppers chose. In many cases, however, the connections it proposes aren't very helpful. Many businesses develop proprietary algorithms to make suggestions for their customers. However, others choose not to take on this task and outsource this activity to firms that specialize in recommendation engine technology. Recommendation engines apply the information collected from user activity to suggest other items for consumption.

The type of items can range from music to television content, videos, books, news, Web pages, and so on. Operating in a business-to-business environment, the companies that develop recommendation engines have to carefully consider the specific needs of their diverse clients in order to be successful. The Filter is a company that thinks it can do a better job for those companies that need this type of functionality, like Yahoo.com. Doug Merrill, current director and a former chief information officer at Google, says, "Recommendations-from friends, from newspapers, from colleagues-are the most common way to find new content. However, there is more information available than there are people to recommend. The Filter analyzes data to provide measurably better, more relevant recommendations, automatically."

Based in Bath, England, The Filter is a privately held company that provides recommendation technology for various companies on the Web. Its founder is Martin Hopkins, a physicist who is also a passionate music fan. In 2004, he started The Filter due to his own frustration as he tried to keep track of his digital music collection of over 10,000 tracks. Hopkins developed an algorithm using artificial intelligence that learned his likes and dislikes and then suggested playlists. This software would later become the foundation for the decision-making procedure that controls The Filter's recommendation engine. The Filter's client list includes firms such as Sony Music, Nokia, and Comcast.

In 2010, The Filter entered into a deal with Dailymotion, a Web site of over 66 million monthly users, to make video-to-video recommendations. During the trial and optimization period, the company's software captured over 5 billion video views and delivered over 1 billion recommendations. Just as Netflix uses its extensive database of customers posted opinions to help locate movies they might not have thought of on their own, The Filter guides customers toward music and video choices that are not obvious.

One of the program's advantages over other recommendation engines is that it only deals with digital media that can be tracked after a customer actually purchases a song online. This advantage allows The Filter to monitor postpurchase consumption behavior. For instance, some customers only listen to or watch parts of the media they download, so this behavior may signal that the customer may not be truly interested in all the material he or she purchased. This type of knowledge could make a significant difference in what the software recommends the next time the user accesses the system. Initially The Filter aimed its services at individual consumers;

however, it did not win enough business to be profitable. As a result, the company shifted direction to a business-to-business environment and began to promote its recommendation engine to media firms, who utilized the service to make recommendations for visitors at their Web sites. The pressure is on to prove that its recommendation algorithms do provide the kind of improvements it claims. The ultimate goal for The Filter is to market its service to companies in industries other than media, entertainment, and technology. The firm has to define its long-term strategy and find ways to ensure profitability well into the future. You Make the Call

1. What is the decision facing The Filter?

2. What factors are important in understanding this decision situation?

3. What are the alternatives?

4. What decision(s) do you recommend?

5. What are some ways to implement your recommendation?