Film Fund-amentals: Data Analytics at the Movies

What do the last election and modern Hollywood have in common?  I mean, aside from ridiculous out-of-control budgets and delusional assumptions.

The answer is data analytics. For example, Mitt Romney really thought he was going to win the election because of his polling data and his use of Orca, the database system that was supposed to do almost everything (including coffee, I suspect). However there were fatal flaws in their methodology on the polling data, and Orca blew up on election day morning.

Not that Orca would have mattered. Everything I have read and heard about the Romney campaign’s approach to data analytics strongly suggests they would have been better off sniffing sulfur fumes and reading goat entrails. They were primarily taking raw data and cherry picking whatever information suited their beliefs and made them look good.

Does this sound familiar? Yes, except we call it show business.

For the film industry, data analytics barely exist. Sure, Hollywood has long been obsessed with focus groups and audience polls. But these days, those approaches are both primitive and extremely limited. Which is why they are not getting any real information and are routinely surprised at box office failure.

But this is changing. Last Wednesday, USC and the Annenberg School for Communication and Journalism together with IBM presented The Power of Crowds: Social Sentiment and the Future of Film, a meeting of data experts and studio honchos. In theory, it was an attempt to create a bridge between data analytics and entertainment production. But if you read the Variety report DailyHeadlines, it sounded a bit more like a polite head-butting contest between science and magic.

OK, we all know the drill. The film industry is composed of a wide range of components that cannot be quantified and reduced to any form of statistical based analysis. That isn’t really true, but it is a deeply held belief in Hollywood. Despite the largely routine and banal predictability of much that happens in the business, it is supposed to be a magically elusive enterprise understood by only a chosen few who have some sort of divinely inspired instinctual knowledge about what the audience wants. Keep that in mind the next time you see the same old crap for the umpteenth time.

Give me a break. A pack of monkeys could do the job of some studio executives. But the magic theory still predominates within the business. There are occasional hints of change and various people within the film industry have explored (and a few have adapted) to new approaches. But the resistance level is high and mostly focused on the concept of social media analysis. For many folks in film, the main use of data analytics is solely to determine if a movie is going to bomb or not. Since this determination is made about a week or two before its opening, it’s a bit like a heart doctor who will only examine the patient while the guy is in the ER with chest pains. Not big on that prevention thing, you know.

Though this resistance is starting to shift, there are still many doubters. Certainly a lot of people in the film industry are convinced that adapting to any form of data analytic system in the production process is going to over ride the necessary creative aspects of the process. Or as Rob Friedman of the Lionsgate Motion Picture Group said: “You will never see us make a movie based on a series of questions posed in a research environment….”  Actually, I am a little surprised to hear Friedman say this because the research polling method is only one small part of the modern equation, and is in need of a major overhaul. But I understand his concern.

The pros and cons were better addressed by the political commentator and author David Sirota in his blog article, Could the Nate Silver Approach Work in Hollywood. As Sirota notes, both the film and music industry are already moving in this direction. So is almost every other form of major business. Heck, even the CIA is working with various database models, and the NSA practically invented the whole thing. In most cases, the results have been good. Data analysis has made possible sustainable models that produce more reliable results.

Of course, Sirota is also correct about the cons. He argues that an excessive application of such approaches would result in a film industry that would increasingly refuse to “…invest in a new wave of genre-busting and paradigm-shifting talents.” The truly non-quantifiable elements of the film making process will indeed suffer if data analytics are applied in a heavy-handed manner without expert input or feedback. A knowledgeable human element is critical to any form of data analysis. Based upon my own experience, an interface between the human and the digital is absolutely required. It is not about choosing one or the other. You need both working together as a system of creative checks and empirical balances. We could call this the Kirk/Spock Synergy theory.

Besides, Sirota’s dire warning about data analytics posing a threat to the riskier but essential nature of the creative process has one slight problem. The drab and banal corporate-produced “cinema of nothing” that he predicates is what we already got. The modern Hollywood system does not need a computer to create mediocre movies. They have been capable of that for years all on their own. The full-blown application of data analytics might just further this grim process. But that has to do with the people using it (which takes us back to the pack-of-monkeys theory). I have no doubt that certain executives within the mainstream film industry would use it as the supreme digital overlord simply because it saves them the time-consuming effort of creating bad movies on their own. Likewise, when the movies still bomb they can then blame the analytics and give themselves a bonus.

But it doesn’t have to be this way. It can it be done differently. It can be done better. It will involve extensive use of data analytics. The systems are ready. We just need to prepare the humans who will be interacting with these systems.