Why did "The Avengers" blow the roof off the box office, while "Battleship" sank to the bottom of the sea? Blame internet chatter. The number of times a film is mentioned in blog posts and social media strongly reflects how much money it is pulling in at the box office, according to a new model developed by Japanese physicists.
The researchers were originally interested in modeling how rumors and word of mouth spread over social networks such as online blogs, Twitter, and Facebook. They focused on the big screen because they learned that daily box office revenue data was available and, "I like movies," says team leader Akira Ishii of Tottori University in Tottori City, Japan. The researchers plotted the number of times a movie was mentioned in blogs and on Facebook pages from several weeks before the movie's release until the end of the first run weeks or months later. They found that the number of mentions on social network sites peaked on opening day and then gradually trailed off, with spikes on weekends. A plot of daily revenue followed a very similar pattern.
Using actual data from "Spider-Man 3," "The Da Vinci Code," and other imported and Japanese movies, the physicists developed a model that takes daily advertising spending and social network data and "can predict the [daily] revenue of the corresponding movie very well," the authors claim in their paper, published online Friday in the New Journal of Physics.
Other groups have tried to analyze the word of mouth effect on a movie's success. Ishii says theirs is the first attempt to include not just direct communication between two individuals but also what they call indirect communications, where someone views a Web page without commenting on it, since websites can record the number of its visitors. Ishii likens this to overhearing a conversation in a cafe. "Indirect communications are very important in explaining real market observations," he says. He thinks his team's work can be extended to online music sales and other consumer purchases. Ishii even successfully used Internet chatter to pick the winners in several recent elections in Japan. But he says he couldn't publicize the results because of legal restrictions on publishing projections during campaigns.
"It is an interesting approach and certainly the use of daily advertising data and the data on blog posts are very innovative aspects of this work," says Sitabhra Sinha, a physicist at the Institute of Mathematical Sciences in Chennai, India, who has applied mathematical techniques to studies of the United States movie market. However, he points out that the model relies on social network data that can't be known in advance. This means that rather than predicting a movie's success, the model is better suited to explaining how its performance evolves over time.
Ishii agrees that his team's model can be use only to extrapolate a sales trend a week or so into the future and does not predict total movie revenue. Still, he thinks it could help fine tune a marketing campaign by, for example, boosting advertising if blog post numbers suggest a film is losing steam.