We have extensively talked about the advantages of offering personalized product recommendations to consumers. These recommendations increase consumers’ decision quality and save them a lot of effort. There are various approaches in personalized recommendation systems either based on past behavior or based on collaborative filtering techniques (see this article on different approaches of recommendation engines).
However, we have to take into account that sometimes personalized recommendations (especially when we passively receive them) feel quite invasive and consumers at some point ignore or even react. Think about the following situation: you visit a website X and look at a specific red t-shirt. You don’t buy it, yet you move continue your browsing. You send a couple of e-mails, you read the news and finally you log in to facebook. On the right banner, you see an ad with this red t-shirt you inspected 1 hour ago. Next morning, you open your computer, you search information about something you saw n TV last night. Google gives you 1million results in less than a second…plus, a red t-shirt on the right banner of Google ads.
This may backfire, not only due to the content (recommend wrong products) but also due to the process of recommending itself (see a recent study from MIT for example). Consumers feel intruded when they have not requested for personalized recommendations, yet they get them based on their behavior. Based on data that they (most of the times) did not actively give permission to access. Not coincidentally, the latest Harvard Business Review (issue Jan-Feb 2012) included a short essay under the title “Stop Collecting Customer Data“.