How To Predict A Box Office Hit

Movie studios spend millions of dollars to track interest in upcoming films. Could Silicon Valley be Hollywood's new oracle?

Movie studios have long tried — with moderate success — to predict which films will be hits. But a new study from Google claims that the search giant has found a way to solve Hollywood's problem.

Earlier this month, Google's research division released a report, "Quantifying Movie Magic with Google Search," that claims to have created several methods for predicting box office revenues at different stages of a movie's release.

Most notably, the report asserts that for films with similar franchise status and seasonality (i.e. if the movie is released during the summer or around a major holiday), search volume for the movie's trailer on YouTube and Google four weeks prior to release can predict opening weekend gross with 94% accuracy.

Google's claims are not the first of their kind; Facebook, Twitter, and even Wikipedia have all presented new ways to predict box office results, though mostly as a result of third-party studies.

Around the time of the Google study, marketing company CitizenNet released an analysis of Facebook data, finding that views and likes to a movie's Facebook page correlated with box office revenue.

But the results were underwhelming — most of the views to a movie's fan page were a result of paid advertisements. If Facebook success is only a proxy for a movie's marketing budget, tracking users' activity wouldn't add much new insight.

Twitter, which has probably received the most attention as a social media crystal ball, is more promising. A 2010 study from HP Labs used tweet volume around the time of a movie's release to generate box office forecasts that were 97.3% accurate. But two years later, Princeton University researchers declared Twitter an unreliable indicator in a paper entitled "Why Watching Movie Tweets Won't Tell The Whole Story?" One issue is that Twitter, which has a user base that's a fraction of Google's and Facebook's, may not provide a representative sample of the entire movie-going public.

A study focusing on Wikipedia was equally inconclusive. The volume of edits to a movie's Wikipedia page did correlate with box office revenue, but was not as accurate for less popular movies.

And none of these models were able to reliably predict results as far in advance as Google's, which reportedly generates opening weekend predictions a month before a film's release.

Why would Hollywood care about this? Movie studios spend millions of dollars on "tracking polls" to gauge movie interest in advance and help to coordinate release dates and marketing schedules. (The major player in this business is a firm called NRG, a company that Slate once called "the secretive research group that helps run the movie business.") This industry, which has grown more competitive in recent years with many newer firms springing up, has historically relied on traditional audience polling as the basis for its predictions. But if social media and search companies have found a better — and cheaper — way to predict a film's opening weekend, movie studios will pay attention.

In many ways, Hollywood is just following the lead of other industries. As was reported in the New York Times Magazine this past weekend, the Obama campaign culled a list of names of "persuadable voters" directly from Facebook, moving away from the logic and instinct that has guided politicos for decades. And Michael Lewis' 2003 book Moneyball showed this change happening in sports, as a small market general manager learned to rely less on the sage wisdom of gut-trusting scouts and more on the scientific rationale of a 27-year old Harvard-educated whiz kid.

By 2011, when the film version of Moneyball was released, the rest of the baseball world had caught up with Billy Beane and the A's. Yet movie marketers working on the film likely resembled the gray-haired scouts more than the quants. With these new developments, it may not be that way for long.

Other insights from the report are below:

Overall search volume is closely linked to box office success.

Unsurprisingly, more movie-related searches on Google correlate with higher ticket sales for the industry as a whole.

Major blockbusters change search behavior.

People search more for specific films around the release of a big movie. When no movie is dominating the box office, people are more likely to make generic searches to discover new movies.

Searches alone predict 70% of the variation in box office performance.

Without controlling for any other variables, Google claims 70% of the variation in box office performance for specific films is represented by normal search volume the week before release.

Ad clicks the week before release are a sign of a bigger opening.

Google estimates that for similar films (those opening in the same number of theaters and either part of or not part of a major franchise), an extra 20,000 paid Google ad clicks can be a sign of an opening weekend that is $7,500,000 better.

Skip to footer