How data drives decision-making at Netflix

We all remember the story about House of Cards – the big gamble that launched what would become a new powerhouse at Netflix: Original Content. The streaming giant purchased 2 full seasons of House of Cards in 2012 – a whopping $100M spend – without seeing a single episode. Instead, they’d looked carefully at data to determine how significant the audience was likely to be.

The company saw that fans of the original UK House of Cards also watched movies that starred Kevin Spacey and were directed by David Fincher, one of the show’s executive producers. Combining these elements with a stellar writing and production team created a recipe for success. The show immediately took off, with some indicating they would maintain their Netflix subscription for the sole purpose of watching House of Cards.

Now a major arm of its business, Netflix Original Content is heavily informed not just by its own user data, but by trends the team sees on social media, viewership in competing markets and predictive algorithms that draw correlations between elements of high-performing content – looking at combinations of talent, storylines, themes and directors.

Driven by data

According to Recode, today Netflix is worth around $140B, following a glowing Q4 earnings report in January that sent the stock soaring 42%. It’s now worth more than McDonald’s and GE, and is creeping in on competitors like Disney and Comcast.

The brand didn’t get here by accident. Every decision made at Netflix is deeply driven by data. According to a presentation by Jeff Magnusson, manager of data platform architecture at Netflix, and engineer Charles Smith, the brand’s data philosophy encompasses 3 key tenets:

1. Data should be accessible, easy to discover, and easy to process for everyone.

2. Whether your dataset is large or small, being able to visualize it makes it easier to explain.

3. The longer you take to find the data, the less valuable it becomes.

The team digs so far into personalized customer data that color breakdowns in cover designs for new original content are determined according to their impact on subscriber viewing habits, recommendations, ratings, and more. Everything is personalized, using their advanced machine learning to offer better recommendations and inform their own upcoming content.

With access to such in-depth data – both from existing customers and from reactions to what competitors are doing in the market – the Netflix team can ask better questions and make informed decisions, without using smaller focus groups and other previous forms of testing during the production stage. Instead they can rely on comprehensive data to inform every decision, large or small.

When they do user testing, the team goes big, often spending millions on a single experiment. According to Megan Imbres, Director/Product Creative at Netflix at the 4As 2017 Strategy Festival, CEO Reed Hastings encourages experimentation where everyone can learn something. “ If you think about it, all the time and energy you’re spending on testing – all the work that’s going into it with our data scientists – everyone is trying to read something. Something that I take to heart working at Netflix is really being able to take these big, aggressive testing swings in order to get that learning, get that hypothesis, and then taper back.”

Rewriting the blockbuster playbook

Their first big original film, Bright, starring Will Smith, cost the company $90M and launched on the platform in December. The purchase of the film concept and subsequent marketing was completely informed by customer data.

Bright saw 11M viewers in its first 3 days, despite decidedly poor reviews from rating sites like Rotten Tomatoes. Ultimately, Netflix was able to rewrite the playbook for blockbuster filmmaking using data.

Netflix began subtly marketing the film to users in March, first categorizing and cataloguing it internally and then using carefully crafted algorithms to display tailored trailers, clips and visuals to individual users over time.

The platform’s internal marketing algorithms allow content like this to have a continual life cycle and reach new viewers over time, as opposed to trying to reach as many as possible on opening weekend as traditional box office premieres must do.

According to The Verge, “The company’s in-house tag phrase for the concept is ‘premiere night is every night.’ For users who don’t know Bright exists, its appearance in the browsing interface will be a moment of discovery, whether it’s December 22nd, 2017, or sometime in 2020.”

Data-driven creative marketing

The Netflix team uses external data inputs like social media to inform where to focus marketing spend and attention. Imbres gave an example of the latest Gilmore Girls season that launched in 2016.

Testing the waters with a few tweets, they saw that anything posted about the show immediately took off virally like wildfire, indicating interest was strong. So they tapped into the excitement on this particular content, taking also to offline tactics like recreating the famous Luke’s Diner in select cities, to make the most out of the season’s release.

Their marketing strategy appears to be working. So much so, in fact, that in addition to increasing spend on content creation, Netflix is focusing dollars heavily on marketing. According to CFO David Wells in Variety, the company plans to increase marketing spending more than 50% in 2018, from $1.3 billion last year to $2 billion this year.

“We used to think every incremental dollar was best spent on content,” but it’s increasing spending on marketing because “we think marketing is a multiplier on the content spend,” Wells said.

According to Wells, the brand is set to spend upwards of $8 billion on content in 2018, with 700 original TV shows and 80 original films set to release globally. At the close of 2017, they had 117.6 million streaming members.