News

Lights! Camera! Action! Data!

research / information systems

Imagine it’s the 1950s, and you’re a kitchen knife salesman. Every morning, after you finish your two eggs with toast, you slip into your ironed slacks, grab your briefcase of knife samples, and hit the pavement.

You work in the daytime because that’s when your customers are most likely to be home. On average, for every 100 doors you knock on, 20 open. Of those, six customers will purchase your knife set. It’s a bit scattershot, but considering what little information you have—simply that daytime gives you the best shot of face time with customers—6% is a terrific return.

This is how, for most of the past century, network television shows came into being. For every 100 scripts a network purchased the rights to produce, an average of 20 were turned into pilots, and six were ultimately broadcast, according to Variety, the weekly entertainment trade magazine.

Just like the knife salesman, the networks “have almost no information about their audience as individuals, who they are or why they watch,” says Michael Smith, a Carnegie Mellon University professor in its Heinz College’s Master of Entertainment Industry Management (MEIM) and Master of Information Systems Management(MISM) programs. Although the networks conduct focus groups and receive basic demographic information and viewership numbers from Nielsen ratings, they’ve never had anywhere near the specificity of information collected by the digital media companies they’re now competing with, says Smith, who was selected to be the Exclusive Data Keynote Speaker at this year’s Sundance Film Festival’s Artist Institute Workshop, where industry experts debate the latest technology, tools, and tactics in social funding, digital distribution, guerilla marketing, and independent theatrical distribution.

And Smith was selected for good reason. In 2012, he and one of his colleagues, MISM professor Rahul Telang, founded a research center within Heinz College called IDEA—the Initiative for Digital Entertainment Analytics—to conduct research on digital content distribution in partnership with some of the biggest players in entertainment, including major movie studios, record labels, publishing houses, and relative newcomers like Amazon and Google.

EVERY TIME YOU STREAM YOUR FAVORITE SHOW, YOU’RE SENDING MOUNTAINS OF DATA TO THE COMPANIES PROVIDING IT TO YOU—NOT JUST WHAT YOU WATCH, BUT WHEN YOU WATCH, HOW LONG YOU WATCH, AND WHAT DEVICE YOU’RE WATCHING IT ON.

Together, they’ve distilled over a decade of research, both their own and the work of others—into a forthcoming book, Streaming, Sharing, Stealing: Big Data and the Future of Entertainment (MIT Press, Fall 2016). The book provides a nuts-and-bolts look at how the streaming companies have risen to prominence, as well as a “perfect storm” of threats currently facing the traditional broadcasters—including piracy, a reluctance to embrace data-driven decision making, and consumers’ movement toward user-generated content.

Director of the MEIM program Dan Green commends the research and acknowledges the growing need for companies in the entertainment industry to embrace new distribution models. “The environment surrounding the industry today is less about distributing through antiquated channels and more about providing individual experiences to consumers at the moment they desire and on devices that are convenient to them,” he says.

If the older network model is a knife salesman playing a numbers game, what are the streaming companies? Simply put: better knife salesmen—not because they’re selling better knives necessarily, but because they have more information than their rivals. Imagine if you gave that same knife salesman a lead that contained the names and addresses of the homeowners who needed a new set of knives, as well as information on what time they’d be home and which specific knives they wanted. That knife salesman would close deal after deal and have more satisfied customers.

Such a lead—customer information—is what digital content distribution companies, the likes of Amazon, Hulu, Netflix, and YouTube, are leveraging over their traditional broadcasting elders such as Disney-ABC and NBCUniversal.

Really, write Smith and Telang, the comparison between traditional media companies and their digital rivals “is a clash between human expertise and data.” Every time you stream your favorite show, you’re sending mountains of data to the companies providing it to you—not just what you watch, but when you watch, how long you watch, and what device you’re watching it on.

Those companies then harness that information using sophisticated analytics to create even more of what you want to watch—and to get you to watch it. Digital content distribution companies, says Telang, “have more information about what’s successful with their audience than their rivals—and they have that information at a really personal level.”

According to Smith and Telang, Netflix’s ability to use that data is part of what’s enabled the company to flourish. Earlier this month, Netflix reported to shareholders that at the end of 2015 the company had more than 74 million members worldwide and is estimated to grow by more than 6 million members in the first quarter of 2016, in part because of its expansion “virtually everywhere but China.” As for earnings, in the 2015 fourth quarter alone, the company reported $1.823 billion in revenue and $43 million in profit.

Add the fact that Netflix led the way for “TV Networks” in nominations for the 2016 Golden Globes—Netflix 8; HBO 7; Starz 6; Amazon Video 5; FX 5; ABC 4; FOX 4; PBS 4; Showtime 3; USA Network 3; AMC 2; The CW 2; BBC America 1; CBS 1; Hulu 1—and it’s no wonder Netflix proclaims on its website: “Internet TV is replacing linear TV.”

Amazon benefits from even more information—purchase histories and searches on other non-video areas of its site. That vast metadata has helped its production wing, Amazon Studios—which began producing television shows in 2013—create “Mozart in the Jungle,” which just won Best Television Series–Comedy or Musical at the 2016 Golden Globes awards (beating shows from HBO, Hulu, and Netflix). Amazon’s CEO, Jeff Bezos, recently declared publicly that he wants an Amazon Studios film to one day win an Oscar—and his company’s ability to harness an incredible wealth of data is why many in the entertainment industry think it’s possible.

Because the digital content streaming companies know their customers more intimately than their more traditional rivals do, they’re better able to do two important things:

  • Create the right content
  • Get you to watch it

One of the clearest examples is highlighted in the introduction of Smith and Telang’s book—how Netflix created its hit political thriller television show, “House of Cards.”

In early 2011, a pitch for “House of Cards” was making the rounds of television networks. The proposed series, essentially an Americanized version of a BBC show, had attracted award-winning director David Fincher (“The Curious Case of Benjamin Button”), Academy Award–winning actor Kevin Spacey (“The Usual Suspects”), and Academy Award–nominated writer Beau Willimon (“The Ides of March”). Despite the A-list talent, networks were hesitant to bite because a political series hadn’t succeeded in network television since “The West Wing” ended in 2006.

Netflix, however, was welcoming, write Smith and Telang: “Ted Sarandos, Netflix’s Chief Content Officer … came to the meeting primarily interested in data—his data—on the individual viewing habits of Netflix’s millions of subscribers.” The data showed that a sizable portion of Netflix’s customers were fans of David Fincher and Kevin Spacey and that many “had rented DVD copies of the original BBC series.”

With that knowledge, Netflix gave “House of Cards” the green light to produce two full seasons, at a cost of $100 million. The show became a hit and the first online-only web television series to receive major Emmy nominations. Now, through three seasons (2013-2015), it has received a total of 33 nominations and six awards in the “Drama Series” category, including the prestigious Outstanding Drama Series, Outstanding Director, Outstanding Lead Actor, and Outstanding Lead Actress.

Netflix has since followed that up with more hit shows such as “Orange Is the New Black” and the documentary “Making a Murderer,” which has recently dominated social media chatter. Basking in such success, the company recently announced plans to nearly double its output of original content in the coming year: 31 titles in contrast to 16 in 2015.

But Smith and Telang say that Netflix’s “true genius” isn’t just in using data to decide which shows to create; the streaming giant relies on research findings to tailor marketing to specific fans. For example, there were nine different “House of Cards” trailers, each emphasizing distinct elements—so Fincher fans saw a different trailer than Spacey fans.

Smith and Telang believe that in the digital world we live in—where every viewer’s attention is pulled in multiple directions by, among others, television shows, movies, video games, and YouTube videos—the ability to market efficiently is data’s true potency. Like a best friend who knows your tastes in and out, streaming companies are using your data information to produce what you want to watch and then using that same direct access to strategically promote the content.

So, why don’t traditional studios employ deep data to make decisions? According to Smith and Telang, they don’t have access to the kind of large-scale data that their digital competition does.

Then, why don’t studios just create their own apps to collect that data? Well, they’re starting to—all of the television networks and many cable channels have launched streaming apps, like WATCH ABC, FOX NOW, and CBS All Access. Premium cable channels like HBO and Showtime—which until 2015 were only available through traditional cable subscriptions—have launched their own direct-subscription “over-the-top” models that allow viewers to stream their content on most devices without a cable subscription. Great, problem solved.

Not so fast, say Smith and Telang. Consumers have shown that they prefer simplicity and want to get as much content from as few outlets as possible. They might balk at learning “how to use multiple websites” and be “unwilling to maintain multiple logins.”

Moreover, the new kids’ information is still more powerful because they compile information from their viewers across all of the content in their libraries—not just a particular network or studio. And they don’t share it. “Amazon, Google, and Netflix …provide no data on customers to their industry partners,” the researchers write.

Why not? Because doing so would help the networks and studios “figure out how much that show’s worth,” says Keith Eich (E’02), a fellow professor at MEIM and former director of digital distribution operations at NBCUniversal. The digital companies pay big money to license the networks’ shows for their vast libraries, he explains; sharing that information would handicap them in negotiations.

So, then, why not team up to create one digital platform that streams multiple networks’ shows? They have—it’s called Hulu, and it’s owned by traditional television titans Disney-ABC, FOX, and NBCUniversal. Partially ad-supported and partially subscription-based, the platform streams television episodes from ABC, CW, FOX, and NBC the day after they air and past seasons of shows, as well as movies.

LIKE A BEST FRIEND WHO KNOWS YOUR TASTES IN AND OUT, STREAMING COMPANIES ARE USING YOUR DATA INFORMATION TO PRODUCE WHAT YOU WANT TO WATCH AND THEN USING THAT SAME DIRECT ACCESS TO STRATEGICALLY PROMOTE THE CONTENT.

The issue, write Smith and Telang, is that Hulu’s success comes at the expense of the shows’ linear presentation—the more people who watch a show via Hulu, the fewer who watch it via their televisions—which cuts into ratings, which is the basis of how much networks charge for ads.

To solve these problems, why don’t the networks fully embrace data-driven decision making? Smith and Telang say that very question was addressed by Richard Hilleman, the chief creative officer for Electronic Arts, during a speaking engagement on the Carnegie Mellon campus. He told students that older companies have always made their decisions “based on someone’s ‘gut feel’ about what will sell in the market,” and those with the best instincts tended to rise to the top of their companies. In contrast, their competition, “Google, Amazon, and Apple … make quantitative decisions based on what their data tells them.”

Still, though, Smith and Telang say that the end is not necessarily nigh for the networks. “We are optimistic about the future of the entertainment industries,” they write—if the traditional media companies “harness the power of detailed customer-level data, and embrace a culture of data-driven decision making.”

But there’s something else on the horizon, something we haven’t touched on yet—the viewing habits of one-fourth of the entire population of the United States—millennials. According to Forbes, the 80 million millennials in America represent “about $200 billion in buying power.” As for their viewing habits, Smith and Telang offer a statistic that might strike fear in television network executives: “TV viewing among 18- to 24-year-olds fell by 32% from 2010 to 2015.” Where are they going? YouTube for one, which, the authors write, “reached more 18- to 34-year-olds” in 2014 “than any cable network.”

That may explain the career path of Andy Forssell, the former CEO of Hulu, who became the COO at Fullscreen this past November—a company that describes itself as “the first media company for the connected generation.”

Fullscreen is a “multi-channel network.” Essentially, it acquires different YouTube channels and connects them with brands and sponsors. “Companies can tell us the audience they want to influence,” says Forssell, who earned a BS in electrical engineering from Carnegie Mellon in 1987, “and we can get them to the right influencers. The companies don’t have to worry about who the influencers are—we have that data.”

Today, Fullscreen reports its 600 million subscribers generate more than 5 billion video views across Fullscreen’s global network each month.

Looking ahead, Forssell thinks that the future of online video entertainment is still very much up in the air. He predicts that the big streaming companies like Amazon, Hulu, and Netflix “will become bigger versions of themselves,”but he doesn’t think the studio system or traditional networks are going to crumble. “There’s a lot of money in that system for a long time to come.” But, he adds, “there are real cracks now.”

Smith, for one, sympathizes with those running the legacy studios and networks, who are “being asked to make billion-dollar decisions without all of the data.” However it plays out, though, he says the “emphasis ought to be on great storytellers telling great stories,” which is a constant in any era.