Big Data Goes to the Movies

If you subscribe to blogs and newsletters in the e-discovery and information governance industries, you have experienced a barrage of advice, webinars and articles about methods and technologies for managing and reducing your business’s Big Data. One company that has become wildly successful in harnessing Big Data is EpagogiX, which sifts through data about thousands of movies to advise the entertainment industry about what scripts will be successful and how a story can be made a box office hit. NPR’s Marketplace reported on EpagogiX in a segment on July 19, 2013. Reporter Stacey Vanek Smith asserts that “[t]he thrill you feel when Brad Pitt faces off with killer zombies in World War Z and the way your heart jumps when Daisy and Gatsby are reunited can be measured and priced.”

EpagogiX analysts have read and coded thousands of scripts according to every facet imaginable—big stars, great plots, action, killings, love scenes, scariness, explosions or humor, to name only a few features that characterize great movies. They use a directory to code these scripts, much as professional on-shore or off-shore document coders use a set of directions to logically unitize or code documents according to specified criteria.

Importantly, the box office success in terms of dollars earned is also captured as a data point. Just as one might filter and associate certain metadata in a litigation document review, producers and directors can filter on, say “thriller,” “scariness,” “female lead” and “dollars earned.” This “objective” data is combined with thousands of viewer surveys the company has conducted—“subjective” data. Feed it into the algorithm, and bingo: Now you can profile what kind of thriller, scariness, explosion or female lead amounts to mega-box office success.

Studios use the data gleaned from these analyses to make critical decisions about expenses such as casting. During one engagement, EpagogiX recommended that a role be scaled back because the data predicted that it didn’t need to be as significant as it was portrayed in the script in order for the film to be successful. The studio executive enthusiastically responded that the advice had saved him $12 million, given that he had originally planned to use a particularly well-known Hollywood star for the role. He could choose a less costly actress and achieve comparable success with the film.

Analysis of the data leads to some interesting conclusions.

If a movie is based on a true story, it’s no longer acceptable to change the facts or sugar-coat the ending. Today’s viewers will Google the story after seeing the movie and won’t like it if the facts have been changed—they expect movies to be factually and historically accurate. For devotees of slasher films, the scarier experiences are those in which the killings are random, rather than being based on a motive such as greed or jealousy.

Entertainment analyst Porter Bibb warns about using Big Data to pick and choose scripts and cast movies. “If you put everything into the meat grinder of big data, you come out with a hugely similar product,” says Bibb. “That is taking away a tremendous amount of creativity and it may, in the end, come back to bite the studios.”

The owners of EpagogiX certainly don’t like to think of their product as a cold, calculating, statistical algorithm, but for now, anyway, they are wildly successful, to the point where certain studios won’t consider a film unless EpagogiX has vetted it.