Hollywood turning to big data to write the next blockbuster

Social media posts and blogs are telling streaming companies and producers what to weave together on-screen for the best chance of bottom-line success. Photo / Getty Images

Your Facebook posts and tweets may contain hidden creativity. In fact, they could be helping to write the next Hollywood blockbuster.

As once-monolithic television audiences splinter and migrate to the internet, viewers have unwittingly turned the creative process upside down. Their social media posts, blogs and file downloads are telling streaming companies and producers what actors, writers and themes to weave together on-screen for the best chance of bottom-line success.

The prize is clear for entertainment companies ranging from Netflix, which has more than 81 million subscribers in 190 countries, to Stan Entertainment, a startup battling the $42 billion Nasdaq-listed rival in Australia. Giving customers what they want -- before they've even asked for it -- increases the odds of a show's success while forging loyalty for the content-provider, industry executives maintain.

"We're trying to get a sense of what people are talking about, what's trending, and what's relevant out there," said Chris Oliver-Taylor, managing director of Matchbox Pictures, the Australian producer of Glitch and The Real Housewives of Melbourne that's owned by NBC Universal Media.

"If the entire Twittersphere is talking about this particular thing, or if there's a trend in this area and we create a piece of content that talks directly to it, then logic suggests that those people will engage with it," Oliver-Taylor said.

The New York-based parent company's leadership conference last year focused almost entirely on how to use big data, he said.

The evolution of home entertainment from free-to-air television to content streamed over the internet to multiple devices has facilitated greater insight into what, when and for how long customers are watching.

It's that knowledge that informed the writing and casting of "Wolf Creek," a new six-part series based on the Australian 2005 horror classic of the same name, according to Stan Entertainment.

"When we sit down with the creators and producers of original productions, we effectively give them a high-level brief of what we're looking for," said Mike Sneesby, Stan's chief executive officer, in an interview. "Part of what's informed that brief is the data that comes from our platform."

For example, Australian actress Lucy Fry, who played the wife of Lee Harvey Oswald in the miniseries "11.22.63", was cast as Wolf Creek's female lead to broaden the show's appeal beyond its slasher origins, Sneesby said. To test if the choice would resonate with women, Stan highlighted Fry's role in 20-second Wolf Creek trailers on social media before the series' release. The results vindicated the decision and indicated the show would work.

Whoever has the most data comes out on top, and the value of metadata will shift from content distribution to actual production, James Sullivan, a managing director of Asian equity research at JPMorgan Chase, said in a report in March.

Still, there's no consensus on the extent to which metadata should influence the entertainment business, he said. Sullivan, who is based in Singapore, recently traveled to North America, Europe and Asia to canvass the opinions of writers, entertainment lawyers, and executives from Google and Netflix on big data.

"The million-dollar question is whether you can build a plot around it," Sullivan said in an interview. "Ultimately, I think the answer will be 'yes,' but it's an incredibly controversial conversation. There have been clear creative concerns expressed to me, particularly by screenwriters."

The million-dollar question is whether you can build a plot around it... Ultimately, I think the answer will be 'yes,' but it's an incredibly controversial conversation.

The entertainment industry is still figuring out how to gain maximum benefit from this high-level customer intelligence, said Matchbox's Oliver-Taylor. Once it has, big-data driven shows will become commonplace, he said.

Knowing what elements could make for a great show isn't enough to guarantee success, Stan's Sneesby said.

"It's not always possible to match what the data says with the availability of projects," said Sneesby, who was in Los Angeles last month for annual screenings by Hollywood studios of pilots and first episodes of upcoming shows.

Amazon.com Inc. says it releases pilots at Amazon Studios periodically for customers to watch and review. Their feedback is taken into account when executives decide which pilots will become a full series.

One product of that system is the comedy series "Transparent," based on a Los Angeles family whose patriarch is transgender. Its debut in 2014 coincided with greater social awareness about transgender issues and was rewarded the following year with the Golden Globe for best TV series, musical or comedy. Star Jeffrey Tambor also won for best actor.

Representatives for Netflix and Amazon said no one was available to discuss the role of big data. A spokesman for HBO, the pay-television of Time Warner Inc., said the company doesn't use data "in any way" to determine its programming schedule.

The art and science is maintaining that sense of wonder while actually understanding what audiences' tastes are.

Netflix, which distributes shows such as "House of Cards" and "Orange Is the New Black," pioneered the use of mathematical equations to promote titles that a subscriber might enjoy. That's based on variables such as previously downloaded content, the subscriber's location and the show's broader popularity.

A typical Netflix user may lose interest unless something interesting is found within 60 seconds, two employees of the Los Gatos, California-based company wrote in a paper published in a scholarly journal last year. Netflix's system for coming up with personalized viewing recommendations helps save more than $1 billion a year by reducing the number of subscription cancellations, they said.

In New Zealand, Parrot Analytics has built a business on measuring and predicting viewer demand. The company, which has offices in Los Angeles and Auckland, says it's analyzed petabytes of content-demand data from consumers in 249 countries to create a platform for gauging and predicting the popularity of shows, including its cast.

"We are in the best position to be able to predict what is happening tomorrow," Parrot CEO Wared Seger said in a phone interview. "Is drama trending up or down in a particular market? Are comedies or horror the next big thing in Japan? Some TV shows are huge in China, but they're not that big in Japan."

Australian public television network SBS attaches dozens of tags to each show to identify such things as its cast, writers, director, themes, plot, location and tone. The information is key to matching viewers with the right content, said Marshall Heald, SBS's director of television and online content.

"Netflix has made this a 'black art,'" Heald said. "We're not at the same level of sophistication, but that's certainly our longer-term aspiration."

Even armed with this intelligence to direct an audience to a show that's bound to appeal, it's important a viewer feels it was their own discovery, he said.

"Often those discoveries feel quite accidental," Heald said. "The art and science is maintaining that sense of wonder while actually understanding what audiences' tastes are."

- Bloomberg.