Blockbuster Inc. has revised the software it uses on its website to convince customers to rent more movies. In the past, Blockbuster.com made recommendations to renters based on obvious criteria. For example, if someone rented a movie with Jack Nicholson in it, Blockbuster would recommend another film with Nicholson. When the site began using its newer software in 2005, it could utilize more subtle factors in making recommendations, such as whether the previous movies rented by an individual were 'mindless' or 'challenging,' or whether they were character-driven or plot-driven. Online purchase recommendations have existed for some time, but they have often resulted in complaints from users because of their simplistic nature or because they were just wrong. Now, websites rely on increasingly sophisticated methods that steer customers to specific products. Blockbuster uses a system developed by ChoiceStream Inc. This works by assigning characteristics to individual items and then searching for other items that have the same characteristics. Combining the data from older techniques for making recommendations, the new system devises a very targeted suggestion. This method is called the 'attributized approach' by ChoiceStream. Improved recommendations have become critical for online retailers as they attempt to keep sales increasing in the face of lower numbers of new online shoppers.