What makes good TV? Apparently the answer is more science than art…

What has historically been the art of the tastemakers in Hollywood is moving to the science of those in Silicon Valley

Netflix has already revolutionized the way entertainment is delivered to consumers but their next goal is even more ambitious.  The company is using big data to guide what content to produce and the user experience.

Historically, content producers like Viacom and Disney chose the shows and movies they would produce based on very little data and would rely on the intuition of executives who were expected to have a pulse on what viewers wanted to watch and what ideas/scripts were good.  Whether they wanted to or not, traditional television broadcasters approached content in this way because they had barely any viewership data outside of estimates produced by Nielsen which were not very accurate or informative [1].  The game changer is that Netflix’s digital distribution network and infrastructure allows it to track every click and every facet of viewership.  Netflix has decided that the art and intuition in the traditional model is less effective than the insights provided by this massive set of data.  Their ability to select popular shows to produce is becoming more important as their industry matures because the price to acquire content made by third parties is rising at an unsustainable rate; Netflix can no longer rely on buying proven shows but has to focus on producing their own original content.  In a few short years, Netflix’s original content production budget has grown to the biggest in the industry,$5B, making selection vital to the business [2].

spending[3]

This strategy has already shown some real success.  Netflix noticed that there was a high level of interest in the show Arrested Development which had been cancelled a few years prior and was originally carried by Fox but was carried by Netflix.  The data said that viewers loved the old episodes and in turn Netflix greenlit a third season of the show as original content for which there was a lot of fanfare.  Their biggest success and bet to date though has been House of Cards.  The show was put out to bid and Netflix had to pay $100M to secure the rights to the show over other top TV networks like AMC and HBO.  This was a big bet to take but Netflix knew from their data how much of a draw this would be because they knew that on Netflix, House of Cards director’s movies were big hits, Kevin Spacey was very popular and the British version which streamed on the platform was also a success [1].  Netflix was able to leverage their data to have a leg up on the competition, like playing a game of poker where everyone else had their cards showing [4].  Finally, as the business goes global they also need to pick content that can be successful across borders.  An example here is that the data showed that Adam Sandler was an actor who had global appeal so they signed him to a 4 picture deal; his first movie became the most watched movie in Netflix history [2].  Ultimately Netflix’s goal is to have enough content that customers want to watch on its platform to attract new customers and keep current ones.  By using data they have figured out a way to more efficiently hit those goals.  What has historically been the art of the tastemakers in Hollywood is moving to the science of those in Silicon Valley.

Not only is data being used to select the content Netflix produces but it is also used to select what content it should serve up to each individual consumer.  They have created algorithms that can help predict what shows or movies a viewer would want to watch based on their tastes, day of the week, time of the day and other factors.  The moment you finish watching something another algorithm can suggest something else to watch immediately when the credits roll based on the prior decision [1]. Aside from catering to your viewing tastes, Netflix even uses data to select the best header images of each show.  For example, when Daredevil was released there were 8 different header images used and based on the click through rate, they selected the most popular one and eliminated the other seven [2].  Similarly, when House of Cards was first released, Netflix produced 10 different trailers and were able to track which one was most effective by tracking who viewed which one and the percent that then went on to watch the show [1].

Netflix’s use of big data, which was made possible by their digital distribution platform, has fundamentally changed the media landscape and helped propel Netflix to great success.  They know what customers want, when they want it and how they want it.  It has allowed them to disrupt the legacy TV business and create a different business model that positions them for success into the future.

[800 words]

[1] https://blog.kissmetrics.com/how-netflix-uses-analytics/

[2] https://www.wired.com/2016/03/netflixs-grand-maybe-crazy-plan-conquer-world/

[3] https://www.bloomberg.com/news/articles/2016-03-02/media-companies-try-to-spend-their-way-out-of-cable-tv-crunch

[4] http://fortune.com/2016/09/19/netflix-streaming-tv-movies/

Previous:

From Digitilization to a New Energy Landscape

Next:

The “Smart House” is on its way.. or is it?

10 thoughts on “What makes good TV? Apparently the answer is more science than art…

  1. Thanks for your post! Netflix’s use of data to outbid competing networks for show rights is fascinating–what a leg up to have! I wonder, though, how Netflix’s use of data to determine what shows to produce will affect the industry more broadly. In the past, knowledgable industry leaders could actually serve as progressive forces to move society forward by taking big bets on new ideas of what society would accept (silly example: in 1967, the same year that the Supreme Court legalized interracial marriage, there was a interracial kiss on Star Trek). HBO’s Six Feet Under featured an interracial homosexual couple in 2001, well before the Supreme Court legalized gay marriage. I have to believe that such inclusion pushes society forward, and the tech-fearful, artist-supporting part of me wonders if using data on past preferences to determine future production will inhibit the representation of new ideas or broader societal trends. TV can be a powerful force for normalizing “new” behavior. But using analytics, will we ultimately end up seeing more of the same tv, or will Netflix be able to use past preference data in creative ways? Will we be getting TV that we like more, but that challenges us less?

  2. MD – It is amazing to see Netflix constantly adapting to maintain its competitive position. When the Company announced its strategy to produce more original content, few envisioned that it would be able to gain this much traction with its viewer base this fast.

    Like the companies we have studied in TOM over the past few weeks, Netflix is also taking steps to increase efficiency in its supply chain. One example of this is through production of its original programming. When the Company first started developing its original content it would source a third-party production studio to produce the series, and ultimately make payment on final delivery. However, as its volume of original media increases, Netflix is now cutting out the middleman and producing original content in-house (i.e. Stranger Things series). Although this requires a larger upfront investment, it benefits from greater control in working with creative teams, lower overhead fees, and the ability to retain Intellectual Property for future uses like toys and games [1]. Given Netflix recent success with in-house production, it should be interesting to see if the Company can continue to maintain its ability to produce products that captivate its customers.

    1. http://www.fool.com/investing/2016/10/20/the-biggest-factor-driving-netflix-subscriber-grow.aspx

  3. It’s a big scary how engineered everyone’s individual Netflix experience is! I’m not entirely sure how I feel about Netflix keeping an eye on which cover image of Unbreakable Kimmy Schmidt is most effective on me…

    But on the other hand, this is nothing new. Digitization offers a faster, more efficient, and far more thorough way for companies like Netflix to keep a tab on consumer preferences, but the music industry has been doing this for a while. Hitmaker lyricists know which formulas to fall back on when penning new songs for the likes of Rihanna and Britney – and as a result, so much of top 40 radio sounds the same (and, the millennial whoop is partly to blame: https://thepatterning.com/2016/08/20/the-millennial-whoop-a-glorious-obsession-with-the-melodic-alternation-between-the-fifth-and-the-third/). Where I think Netflix’s handle on data will provide the most value is in the ability of their algorithms to identify very minor or almost imperceptible consumer preferences (the beginnings of a new trend) and invest in innovative, creative content that breaks norms away from existing mainstream content but gets credit for tapping into yet-unrealized content preferences.

    Or, you know, I’d be happy with more Arrested Development, too.

  4. Thanks MD. It is indeed an interesting blog on how technology is transforming Netflix’s operating model and how Netflix is taking advantage of big data to know what customers want and deliver their content accordingly. However, I am wondering whether using big data to understand customer’s preference could hurt Netflix in the production phase. I agree with the firm’s strategy to use big data for marketing and distribution, but for production of the content, I am afraid that Netflix would be constraining the content to what the data shows what we like. This might impede on the creativity of the production and might as well miss out on things beyond the data i.e. things that our behavior doesn’t translates into what we like or don’t like. Our behavior may be misleading sometimes and we might not know what we like until we actually watch something. I would go beyond that and point out that consumption may not be correlated with satisfaction, and although we might consumer something over the short-term e.g. an Adam Sandler movie, if we are disappointed by this movie, this may lead us not to watch the next one, so a 4-movie bet, may be a great bet. This is what’s happening right now, Adam Sandler’s first movie has a 0% rating and a 31% audience approval on Rotten Tomatoes. His second movie also has low ratings. So would a Netflix user watch the third movie when he or she who has been disappointed twice? These are a couple of articles worth looking at: http://www.grunge.com/12267/netflix-lost-adam-sandler-movie-deal/ http://www.indiewire.com/2016/05/the-do-over-review-adam-sandlers-awful-new-netflix-comedy-will-make-you-wish-hed-stop-trying-510400/

  5. Great post, I really enjoyed reading this- I love all things Netflix, and am a huge fan!
    I have some concerns regarding their growth trajectory and how it might affect the neutrality/fairness of their big data use and delivery to customers. A huge part of Netflix’s business model comes from other content providers that are trying their very best to upsell their own content streaming platforms, and as Netflix expands in terms of both geographies and inventory, and becomes more popular, content deals are becoming more and more expensive. I wonder how the need to pacify and maintain relationships with large scale media providers might influence the way Netflix uses/or does not use its data to influence viewership. What’s your take on this?

  6. Great post, MD. It is very interesting to see how Netflix has made such a big chance in the way content is produced and in the way the content managers base their intuitions on data. As a Netflix power user, I have noticed how they A/B test the pictures that they use for the same movie.

    The only drawback that I would see in their strategy is the fact that the diversity of the content on the platform might suffer. The creativity of the content might be hindered by the fact that Netflix would focus their production efforts on the content that they know will work and might become very risk-averse in the content they buy. The Netflix business model so far relied on the diversity of their content and I am afraid that they might actually become a “new network” and only produce/buy mainstream content.

  7. At some point, I actually went through the 2009 $1m award winning algorithm that Netflix never implemented. Here’s my take on this though. Firstly, given the fact that no Netflix viewership numbers are ever disclosed, I would be very wary of calling anything a success on Netflix, even House of Cards. Sometimes, we choose to forget the fact that we hang out socially with a lot of people very much like us and assume that if 4 out of the 5 people have watched a particular show, then 80% of the globe must have watched it. Nothing could be farther from the truth. Second, Netflix, as it steps into the playing field of “feeding the monster” as they call it, has begun to realize the amount of content, just in pure play volume required to sustain customer interest. A look at the Netflix home page clarifies it, when they start with XX new shows and YY new movies added in the last week. And as much as we love to rant about HoC and Narcos, let us not forget that these are 2 albeit 3 shows that have caught universal attention. Thirdly, on the usage of data finally, Netflix realized long back that the 86000 or so genres that they had floating around were more of a distraction than enablers, as people spent hours browsing through the recommended lists, rather than actually watching content. With respect to the header images, their research told them what was so much more appreciated was a single individual on the snapshot, instead of multiplicity of characters, so they implemented it – guess what the home page looks like now – like a photo diary!

  8. MD! Great post on one of my favorite products. It’s very interesting to read how Netflix leverages data to determine what content to purchase and produce. In my view, the data-centric method for content development has really cemented Netflix’s spot as the top streaming service. Initially it was about access to movies and TV shows, but I believe the tables have turned and original content has become the biggest driver of views. With Amazon and other competitors leveraging data to buy the right content, what will Netflix do next to stay ahead of the competition?

  9. Great post, I really enjoyed learning about the science behind the content. I would also love to hear more about how this technology actually works, as it seems to be a fairly complex system that must take some serious brainpower to operate. I also am very interested in what they come up with next, because they really do not seem to ever put out a show that flops. Finally it will be very interesting to see if Netflix can ever actually justify their outrageous valuation even though they lose money at a rapid clip, I for one have bet big on them. I feel that it is still a risky bet though, unless they can keep adding user at a rapid pace and finally achieve the necessary scale.

  10. Great job, Michael. Think you went a little long though. Plugged this into Microsoft Word and it looks like its 821 words. Just something maybe worth double checking.

    Super interesting post. It seems that we are getting way too caught up in outsourcing “big data” to make all of our decisions, but even the most sophisticated analytics operations are using to both guide and supplement human thought. Arrested Development being the prime example of a show that had success after it was canceled that Netflix was both able to A) recognize the creative genius and B) prove there was market demand given all of these online signals. Really cool that a company that essentially started as a glorified supply chain for DVD’s now has this internal capability.

Leave a comment