Netflix: Your Data, Your Show, Your Experience
Netflix has been the poster child of combining data-driven growth with creative choices. At its core, the Netflix big data approach to media has resulted in more personalized entertainment experiences for subscribers and better creative decision making.
Netflix has long prided itself on the quantitative approach it takes to providing consumers with a world-class experience. The results of the big data strategy are impressive – Netflix’s ~120 million subscribers watch over 140 million hours of video every day on over 450 million different devices (1, 2). The Netflix analytics software and recommendation engine are at the heart of what makes the platform so effective. In the words of Joris Evers, Netflix Director of Global Communications, “there are 33 million different versions of Netflix” (3).
Netflix’s core competency in data science enables the personalization of the streaming experience based on user behavior. Netflix classifies and tags content to get a nuanced view of consumer preferences. Netflix has developed over 1,000 tag types that classify content by genre, time period, plot conclusiveness, mood, etc. These tags help to define micro-genres, which, by 2014, had already reached 76,897 (4). Content micro-classification, combined with a proprietary recommendation engine, enables Netflix to serve customers better. About 75% – 80% of viewer activity is influenced by the recommendation algorithm (3). The true north of the Netflix big data strategy is the philosophy of best serving specific audiences rather than “being all things to all people,” a mantra of old broadcast TV shows (5). Utilizing data has enabled Netflix to execute their philosophy for each user, demonstrating to traditionalists that melding data with creative intuition can produce superior performance.
The other major advantage data affords is improved creative decisions. Metrics like completion rate, stop and start time, time of day, and viewing behavior (e.g., pause, fast forward, rewind, etc.) allow Netflix to make better content decisions. When greenlighting Arrested Development, Netflix saw how many users, of those who watched through to Season 3, completed the series. This helped producers decide not only whether or not to greenlight additional runs, but also what particularly resonated with audiences given audience behavior when watching certain episodes. The classic example of Netflix’s prowess was House of Cards, which rose out of a confluence of a) data on the popularity of the British version of the show, b) fans of British House of Cards who also watched a lot of movies that involved Kevin Spacey or were directed by David Fincher, and c) those who watched David Fincher movies over indexed on completion in a single sitting (3). The result? Success rates for Netflix shows are 80% vs. an average of 35% for traditional TV shows (6). Netflix has established the industry standard of data and creative talent. HBR researchers have shown that a data-driven approach to a creative endeavor, like producing TV shows, has led to greater product variety for audiences (5). As Netflix’s subscriber base grows, so does its defensible moat. The sheer volume of data is itself a competitive advantage. Some observers note that it took 6 years before Netflix reached the point they had enough data to help create shows from scratch (6). In greenlighting Orange is the New Black, Netflix had already determined the likelihood of success based on viewership data of women-led TV shows and Jenji Johan’s hit show Weeds – Netflix knew exactly which subscribers would be interested, reducing the creative risk of the new show (5). Of course, analytics does not solely drive creative decisions; Netflix still involves production studios and creative staff to make shows a reality. However, Netflix pioneered the art/science approach of data in entertainment. The natural question of how the entrance of Amazon and, to a lesser degree, Hulu impact Netflix’s future. But that’s a topic for another post.
References
- Netflix Subscribers Streamed Record-Breaking 350 Million Hours of Video on Jan. 7 [Internet].; 2018 [updated March 8,; cited 4/4/2018]. Available from: http://variety.com/2018/digital/news/netflix-350-million-hours-1202721679/.
- Netflix now has nearly 118 million streaming subscribers globally [Internet].; 2018 [updated Jan 22,; ]. Available from: https://www.recode.net/2018/1/22/16920150/netflix-q4-2017-earnings-subscribers.
- How Netflix Uses Analytics To Select Movies, Create Content, and Make Multimillion Dollar Decisions [Internet]. []. Available from: https://blog.kissmetrics.com/how-netflix-uses-analytics/.
- Jenkins T. Netflix’s geek-chic: how one company leveraged its big data to change the entertainment industry. Jump Cut. 2016 Oct 1,(57):Web.
- Smith M, Telang R. Data Can Enhance Creative Projects — Just Look at Netflix. Harvard Business Review. 2018 January 23,.
- How Netflix Uses Big Data
[Internet].; 2018 [updated Jan 12; ]. Available from: https://medium.com/swlh/how-netflix-uses-big-data-20b5419c1edf.
Great post Sean! Netflix’s creation of proprietary data (e.g., through the 1000 tags mentioned above) is an interesting complement to its customer generated data and opens up really interesting possibilities for machine learning enabled analyses — it is fascinating that they have used these data to identify and capitalize on the micro-genres and micro-segments mentioned. The outcomes for their creative activities (80% vs traditional 35%) are incredible! It would be very interesting to know how this compares to other viewing platforms trying to get into the content creation business (e.g., Amazon) and what the competitive response from traditional content providers will likely be. Given the amount of data Netflix has generated, and is generating everyday, is it possible that they have they built a sustainable moat in terms of their data and algorithms?
Thanks, Sean. I thought it was particularly interesting how Netflix used the data in its production of TV shows. I wonder how other TV and movie producers use data in their productions. I’d imagine it’s a very controversial topic in the industry, with many people unhappy that the artistic nature of their work is being dampened by the numbers. They also have to think about varying goals: box office sales, award shows, ability to sell in international markets, and ability to produce sequels.
Great post, Sean, thanks! I agree with all of your thoughts and comments about Netflix and how they have been utilising data to improve the offerings it gives to its customers, but I wonder then how they decide to stop hosting shows or films that are very popular. Do you think that it is to make people feel excitement when they return? Or perhaps due to licensing deals? I switched to Amazon Prime precisely because Netflix stopped offering the show that I wanted to watch and Amazon had it for free. It is interesting to me that Netflix did not pick this up in its data and try to mitigate this as it has happened several times since then.
Thank you for your post Sean. It was very interesting to know more about how data is used in the creative process of Netflix new shows!
Reading your post, similarly to Zach, I got worried about the creative process of new shows and how data might affect it. In my opinion, it would be detrimental to users if this approach ultimately led to a convergence of “blockbuster” themes that appeal to large audiences and are likely to be successful, as opposed to new, more creative ideas. At the same time, I wonder if the fact that Netflix knows which subscribers are likely to be interested in a particular show, thus reducing the risk, as you mentioned, could be used as an advantage. As micro-classifications and likely audiences are identified, could people be incentivized to produce targeted content by investing small amounts that make economic sense, hence increasing the amount of new and differentiated content?
I think a lot of what is in this post is what Netflix really wants consumers to think. I would, however, point out a few things that show that even Netflix does not use a pure data driven approach. First, when it comes to its recommendation engine, it is clear that Netflix manipulates this beyond pure data analytics. The company will often promote its own shows and downplay shows or movies it has less confidence in (or similar offerings of licensed content) to the detriment of consumers. It’s not purely data based. Second, when it comes to making content, I would argue that Netflix is indeed trying to be everything for everyone. The sheer volume of spend and content indicates that is trying to appeal to everyone, not specific audiences. I liken its content strategy to that of dredging the ocean floor and seeing what pops up and works. For every House of Cards, there are a ton of other shows that we don’t know about because they simply did not work – like all other creative endeavors. Indeed, Netflix is now much quicker to cancel shows, understanding that data-driven creativity does not always work. Despite my pessimism, it is true that in the world of entertainment, Netflix is certainly leading the way in how it uses data.
Thanks Sean, great post. I do agree that Netflix is definitely the industry leader when it comes to data usage. Having said that, from a user standpoint, I still think they have a long way to go. The sheer volume of their library means that discovering new movies/shows is extremely difficult. I wonder how Netflix can better optimize this process or if this should be their focus at all.
Thanks, Sean! It’s interesting to see the value of Netflix’s data. I’ve often heard it criticized for not creating very useful suggestions for viewers on the app. But perhaps that’s missing the point. Perhaps the real value is what you underline here regarding the use of data to green light shows more accurately, and to develop new content to serve micro markets among viewers. That certainly seems more believable to me than the thumbs up thumbs down matching system one sees as a user. I often feel Netflix pigeonholes me into seeing the same types of shows I’ve seen before, and actually prevents me from exploring and enjoying the larger library of content. Thanks again!
Thanks for an interesting read Sean! I agree with you that Netflix is pioneering in using art/science approach in content creation. Amazon tried to produce a similar themed show to House of Card, Alpha House, but only saw it achieve 7.1 ratings (vs 9.1 rating of House of Card). Some analyzed it’s because Amazon tried to use data all the way from sourcing ideas and creating plot, while Netflix approach is data-driven on initial idea-sourcing but more art/brain-driven in concept and plot creation phase. In general, the ability to create successful original content is vital especially when Netflix and Amazon video tries to expand internationally, because international content rights are hideously complex and having your own exclusive IP saves the headache.
Thanks for posting Sean! I have to say I am a little skeptical about the use of data for what I see as a very creative product. I know that data informed decisions can do things like greenlight Orange is the New Black, but I fear that if you use data to create a formulaic production, it won’t have the same heart.
And I also wonder if using data to see what’s gaining traction is really a new concept. There have always been ratings and ways to measure viewership. Of course the time something was showing played a factor, but it’s not like we were operating in a world without data that now has come to light. I just wonder how valuable it really is, and if you couldn’t make similar recommendations without all of this data jsut knowing basic demographics of your users. Realistically you don’t want it to be TOO personalized because you still want to allow for serendipity/discovery.
Thanks for a great post! I completely agree that the smart application of big data has been Netflix’s winning strategy. However, as more players are aware of and equipped with data analytics capabilities, I think the accuracy of the data insights would become the differentiating factor. I would be curious to see how Netflix is going to sustain its advantage over time.
I really enjoyed this article. I wonder if too much data my steer Netflix the wrong way as it looks to grow its business and expand into different markets. I absolutely love what they are doing, especially in terms of their geographic diversification strategy, and think there are clear values that are captured by both the company and the subscribers. I have however witnessed a decline in customer engagement and interface improvements. More data provides the company with more opportunity, but as it grows perhaps Netflix should consider separating its production units. I also believe there is a myriad of different factors that contribute to the success of a movie, so I wonder how useful the data is with time as the creative/human aspect of production decreases and ideas narrow in to similar categories.
Thanks for your thoughtful and valuable contribution, Sean! Your description of Netflix’s recommendation engine reminds me of the perennial dilemma on the “echo chamber” effect of information retrieval and aggregation systems. Recommending similar digital contents may indeed drive engagement metrics in the short run, but may also shield the users from discovering thrilling new content like they’ve never seen before. Going forward, how can Netflix leverage its data analytics capability to strike the fine balance between content relevance vs. content discovery?
Great post Sean! I think your point that “analytics does not solely drive creative decisions; Netflix still involves production studios and creative staff to make shows a reality” is particularly interesting and a concern for me as I think about the company’s future. Do you think there will ever be a point where Netflix becomes more like a movie studio with a core competency in data-driven decision making as opposed to a personalized streaming platform that also produces shows? I’m not sure if the company is already doing this now, but I foresee Netflix striking exclusive development deals with smaller studios (maybe the likes of A24 or Annapurna Pictures), or even outright acquisition of a studio.