Data drives the decisions behind making movies

Movies don't get made without data, and lots of it.

While analytics may not be the first thing that comes to mind when imagining the glamour of Hollywood, data drives decisions at every step of the filmmaking process.

Analytics help determine which scripts to greenlight, who's chosen to direct, which actors to cast in a given role and the diversity of the cast as a whole, where to shoot, the order to shoot scenes and how to tailor different cuts of the film to different audiences -- such as in the United States, Europe, Asia, for airlines, or in director's cut or DVD format. Analytics also drive decisions about how to market the film, which cities and theaters within those cities to target, and other promotion strategies.

On June 11, Tamr, a data curation vendor founded in 2012 and based in Cambridge, Mass., presented a webinar, Connecting Data: A True Hollywood Love Story, featuring Eric Iverson, president, CIO and CTO of Iverson Consulting. Iverson was formerly the CIO and CTO of Creative Artists Agency (CAA) and CIO of Sony Pictures Television.

"A lot of people think it's super fluffy and it's all creative and things just kind of happen and there's no science to it," Iverson said. "Friends of mine who work in healthcare or the sciences, they just can't imagine there's a science behind it. But the truth is, media entertainment is actually a very complicated business, and it's hard."

Despite the piles of money generated by many films and televisions series, the margins between the cost of making them and their gross income are often quite small.

Some make substantial profits in the end, but many movies fail to recoup the cost it took to make them.

And that makes analytics, the ability to predict as well as possible, critical.

"You have to think really long and hard about what you're actually going to make, trying to find audiences and trying to get the right ingredients together to get stories that really work," Iverson said.

Hollywood, Iverson pointed out, in many ways is not different than any industry. Just like anything else that results in a finished product to sell, it relies on a supply chain, and data is behind most decisions along that supply chain in order to optimize the final result.

The first part of the chain is the preproduction, a process often so complicated that entire businesses are dedicated just to the financing and deal-making that go into getting the moviemaking process started.

Next comes the production itself, the studios that do everything from finalizing a script to casting a movie to filming and editing it.

Friends of mine who work in healthcare or the sciences, they just can't imagine there's a science behind it. But the truth is, media entertainment is actually a very complicated business, and it's hard.

One of the key data points film production organizations use toward the end of that stage is the release of the trailer.

Once that's released onto YouTube, studios can track the number of views, how many likes it gets, what the comments say, whether people stop it or watch it all the way through, and even the demographics of the people clicking on it. That informs the studios whether they might need to make changes to the film before it's ultimately released.

When a movie is nearly ready, organizations responsible for distributing the content are brought into the mix.

And finally, the last link in the supply chain is the venue where the finished product is displayed, the theaters and streaming services.

"That is going to create a lot of data," Iverson said. "And a lot of data is going to be required in each of these steps to make the decisions and move things forward in the workflow."

One of the more important results of analytics in Hollywood is the relatively recent realization that more diversity in a given production leads to a more diverse -- and bigger -- audience. CAA went back through the history of film and looked at what effect a diverse cast had on performance at the box office.

"What we found is, over and again, when you had more diverse casts, whether it's by gender, whether it's by race, sexual orientation, you get more audience because that's more reflective of who we are," Iverson said. "Being more inclusive is actually good for us, and we've found that 100% factually true when it comes to audience analytics. We used connected data to help connect us more."

Hollywood's data, however, hasn't always been connected.

As data has become both more available given increased sources such as streaming platforms and become more complex, like any industry dealing with big data, one of the challenges the movie business has faced is disorganized data.

And Hollywood has dealt with it no differently than any other business.

Like organizations in most industries, the various companies involved in the entertainment industry have had to overcome siloed data.

Different companies, whether the studios themselves or any other firm involved in entertainment, kept their information secret. They all had data about online customers, in-person and European and American customers, data from public and private sources and data about licensing and broadcasting rights.

They just threw all that information into a data warehouse or data lake and left it there.

The studios and other moviemaking businesses had to clean up the data mess.

"Tracking of these workflows became really slow at the same time we thought it should be picking up speed," Iverson said. "As we gather more data, if it is not organized, it actually slows down our ability to have effective workflows."

With the help of machine learning to speed up the daunting process of organizing decades of data, data operations teams got the monumental task done. The data people automated more workflows, and eventually, the various companies involved in making movies -- and music, television and any other medium that makes up the entertainment industry -- decided their data should be connected.

"Connected data is autonomous statistics or facts that are joined together -- they're combined and linked -- and by doing that they have the potential to produce exponentially more meaningful information [than disconnected data]," Iverson said. "And that can drive insights and experiences."

Whether making movies or any product.