Tweets Predict Box Office Success
by Leslie Meredith, TechNewsDaily Senior Writer
August 02 2012 08:01 PM ET
More money for more tweets doesn't mean success, according to new HP study.
CREDIT: Twitter: James Cameron (avatar_movie)
Can Twitter predict the success of movies at the box office? HP Labs in Palo Alto, Calif., says it's come up with a simple method to forecast movie ticket revenues by analyzing chatter on Twitter.
And they're optimistic they can apply their method to forecast more important events, such as this year's presidential election.
HP Labs used data from 24 movies released in 2009 and 2010, including blockbusters "Avatar" and "Twilight: New Moon." In all, they considered 2.89 million tweets from 1.2 million users.
To get a feeling for how many tweets make for success or failure, consider that "Avatar" averaged more than 1,200 tweets an hour in its pre-release week and generated $77 million its first weekend, "New Moon" had 1,366 tweets an hour and raked in $142 million, while "Transylmania" had the lowest tweet rate at 2.75 and made about $264,000.
Promotional tweets made by studios and their public relations people (distinguished by the inclusion of links to trailers and promo sites) were of little use in predicting a movie's success at the box office — probably not what studios want to hear. Big promotional budgets on Twitter don't translate into ticket sales, the study concluded.
However, the number of tweets made by ordinary Twitter users in the week before a movie's release predicted opening weekend sales pretty accurately — and better than other measures, such as movie exchange sites. Here participants buy and sell shares with virtual money to determine a movie's worth, similar to the way a real-world stock exchange determines share prices.
The study compared prices from the Hollywood Stock Exchange (HSX), which can be considered the gold standard for the movie industry, according to HP Labs.
"The model built using the tweet rate outperforms the HSX-based model," HP Labs said. "The fact that a simple mathematical model considering only the rate of tweets on movies can perform better than artiﬁcial money markets, illustrates the power of social media."
But the story doesn't end on opening weekend. Researchers also wanted to know if people's reactions to a film that they posted on Twitter would be a better indicator of a movie's longer-term success than a simple tweet rate rate. While measuring the sentiment — positive, negative or neutral — improved revenue forecasts a bit, tweet rate was far more telling.
HP Labs has high hopes for their Twitter-tracking method and says it can be applied to a large range of topics, including as an accurate indicator of election outcomes.
Twitter itself has been taking a stab at that, introducing a new Political Index this week showing how Obama and Romney are faring in their battle for the presidency. The Twitter index does analyze "sentiment," however, and the HP Labs findings might provide a cautionary tale.
Then there was the epic failure of analytics firm SocialMatica in predicting Super Tuesday primary results. SocialMatica measured both the number and sentiments of posts on Twitter, Facebook and other social media sites and called almost every primary wrong. Perhaps Twitter users are more likely to follow through on seeing a movie than they are on making it to the polls.