In Hollywood, buzz beats star power when it comes to predicting box office take
Movie studios love awards season. Winning one of the glittery statuettes that are annually bestowed upon those in the biz can provide a hefty box office boost. But if you are going to put money on which movies will sell the most tickets in the long run, accolades from critics and peers aren’t a very good crystal ball. When it comes to predicting box office success, it turns out that the little people really do matter.
Hollywood’s conventional wisdom says a picture’s success depends on intrinsic qualities like a big-name star, a struggling-hero narrative or great special effects. But if you’re trying to predict whether a film will be a blockbuster — and predict it early in the game — those sorts of things can lead you astray, says Richard Colbaugh of Sandia National Laboratories in Albuquerque.
“As fun as it is to look at George Clooney, that’s not very predictive,” Colbaugh says.
A second common misperception that studios have — and one that’s pervasive in marketing consumer goods — is that influential people will, well, influence things. By that thinking, if a studio wants a movie to be a roaring success it should try to get Oprah to sing the film’s praises.
Now, Oprah may like a lot of successful films, and George Clooney may have been in some of them. But using such metrics to predict whether a movie will be a blockbuster doesn’t work, Colbaugh and his Sandia colleague Kristin Glass discovered.
The researchers started by considering more than 300 films that came out in 2010. They gathered data from Google Trends on how many times a particular film was searched for in the month before a movie was released. They also looked at the Wikipedia entries for each film (many of which are created well before a movie comes out) and how many different people were looking at and editing each film’s page. Then they fed a computer program all these data, along with the factors typically linked to success, such as the starring actors or the producer.
This unconventional approach revealed that predicting box office success is all about early buzz. The number of people looking for information about a movie — regardless of whether the commentary was positive or negative — and the geographic distribution of that activity was most predictive of box office dominance. “Getting buzz going is more important than genre, than narrative arc, than the producer and George Clooney,” Colbaugh says. “There’s a snowball effect. A little advantage in the very early days gets magnified and amplified and then you’re golden.”
But the buzz doesn’t help much if it’s only in Los Angeles or New York. Also important is where the buzz is distributed. If people are googling a film in Manhattan, N.Y., and Manhattan, Kan., as well as Dallas, Seattle and elsewhere, it is a much better predictor of success than a high volume of chatter concentrated in only a few places.
Zooming in on the distribution of early buzz allowed the researchers to predict whether a film was going to be a financial success or a flop a month before the film even came out, they reported in December at the winter simulation conference in Berlin.
Tapping into early buzz also allowed Colbaugh and his colleagues to predict future success in another famously unpredictable sector of the entertainment business. In 2006 Duncan Watts, a network theory scientist who now works at Yahoo!, and his colleagues published a Science paper describing experiments that illustrated the capriciousness of pop music success.
The researchers recruited more than 14,000 people to participate in an artificial music market: When presented with a list of 48 songs by unknown bands, each chose songs to listen to, rated them and was given the option of downloading the songs. Some of the participants were shown only the name of the song and the band. Others were subjected to the experiment’s equivalent of buzz: They could see how many times the song had been downloaded by others in their particular group.
In the initial experiments, including download data made it much harder to predict the success of a song. But when Colbaugh and his colleagues later applied their distributed buzz approach to the music market data, they succeeded in predicting success. Songs that were downloaded a lot early on ended up dominating in those study groups where participants could see what had been downloaded.
Movie studios typically reserve part of a film’s marketing budget for spending after the picture’s release, says Colbaugh. This money is often spent campaigning for awards. But seeding chatter in widely distributed cities might better serve the bottom line, he says.
So on Oscar night, when Hollywood’s most bankable stars are thanking each other and their mothers for their success, maybe it wouldn’t be so inappropriate for them to take a moment and thank the real source of their good fortune — the little people.
SN Prime | February 4, 2013 | Vol. 3, No. 5
