All Stories

Don’t Do That/Do This

Making all Marketing Personal.

“If you liked X, then you’ll love Y!” Online companies like Netflix and Amazon use this personalized recommender system to promote products to consumers with varying degrees of accuracy. Nachiketa Sahoo, assistant professor of information systems, has developed a way to improve the system—and help you create smarter promotions.

Don’t Do That Many companies will survey consumers’ preferences for only a few months, then use that data to recommend products that might appeal to them now. They figure that if consumers’ tastes change over time, then data about their prior preferences is irrelevant.

Do This “Don’t throw out old data, but integrate it intelligently,” Sahoo says. “Though a consumer’s tastes might have changed, his or her previous tastes might be useful in making recommendations to other consumers. For example, I had certain preferences when I was 15. Though I’m 30 now, the data about my old preferences could be used to make recommendations to consumers who are currently 15. Also, based on what we know about an individual, we can predict what he or she might like in the future. Instead of making recommendations based on who consumers are now, try to figure out who they are going to become.”