But behind the billion-dollar success is a hard-working analytics team who review everything from subscriber trends, to viewership, and production data.
“Most people know about the data science of Netflix as equivalent to the recommendation work,” says senior data scientist Laurence Wong. “But the company’s strategy has changed and we’re now currently not just a distributor of content, but also a creator and producer of our own content as well.”
Netflix is always looking for that perfect harmony between data and human judgment.”
The introduction of original content to Netflix’s library has introduced new challenges for Wong’s Content Science and Algorithms Team, which, unsurprisingly, tackles content-related problems.
“That includes forecast and demand, analytics – for the whole suite of programming we have, not just at the show level – and a lot more, but my critical focus is original programming.”
Wong, a Stanford-educated economist with a Ph.D. in econometrics, likes to think of his role as ‘bringing data science to Hollywood.’
“What we mean by that is we’re moving very quickly from being just a distributor of content to also being a creator. Now that we’re venturing into that space, we want to first and foremost understand what that creative process is like and understand how existing Hollywood studios do it. But at the same time, we don’t want to do things the same old way, we want to do things our way as well, and that’s combining [the creative process] with a tech- and data-driven mindset and hopefully invent something new; an entirely different sort of process.”
Wong is quick to point out that in Hollywood, lots of decisions in the industry are made based on the creative instinct. “We want to preserve that. A lot of time, people think that a tech company just wants to replace everything with machines and whatnot. That’s not the case here. If you think of filmmaking as an endeavour, ultimately it’s a creative process. It would be silly to think that that could be completely replaced by, say, machine intelligence. But at the same time, we don’t want to ignore what we can learn from data and statistics.”
Netflix, Wong explains, is always looking for that perfect harmony between data and human judgment.
One example Wong gives of this balance between the creative and the analytical is Netflix’s hit series Orange Is The New Black (OITNB). “It’s a show about a rather unusual concept – a women’s prison – and it’s not in a particularly ‘hot’ genre. If you think about using models based on past data, it’s hard to be bullish on something like this because it’s so unusual.”
In this case, Netflix focused a lot on the creative side of things. “The creative was really, really good,” says Wong. “If you’re thinking about the break-out shows, often they’re extremely different from what is conventional, what’s already existing. So you make the bet using the data to quantify the risk.”
You trust the data, but you also give maximum creative freedom to elevate the genre and storytelling and create something novel.”
On the opposite end of the spectrum is Daredevil, another original program. “The superhero genre is super-hot right now, so even shows that I don’t think are particularly good are doing really well. In this case, you trust the data, but you also give maximum creative freedom to elevate the genre and storytelling and create something novel.”
While the superhero genre has a lot of on-site data and models to guide Netflix through the creative process, when novel concept shows like OITNB come along, Wong analyses other sources, like quality of scripts, projected finances and budgets, and talent modeling.
“If [someone] comes to us and says, ‘I want to create this show,’ what can we infer about the probability of success based on who that person is? If we know that Kevin Spacey is on board for House of Cards, then we would update our probability of success to be higher. So that kind of thinking, that kind of modeling of talent caliber, is something that we make use of.”
But perhaps the best part of Netflix’s strategy is how they measure success. “We validate success holistically,” says Wong. “We don’t just look at if the show is being watched, we also look at how they’re watching it. Are they binge-watching, or watching it on a schedule?”
But this is only one metric, he says. Outside of this data is social buzz, whether it’s turning into a cultural phenomenon, like Game of Thrones, as well as its potential for awards and award nominations. The creative side isn’t lost to the numbers, he adds.
“There have been critically acclaimed shows without a huge viewership,” says Wong. “We don’t need a huge audience for every show. We just strive to have a show for everyone.”
Originally posted on BeMyApp Media This article was based on Laurence Wong’s talk for With The Best‘s web conference on AI.