Netflix’s distribution channels

Having spent enough time on the cost side of Netflix, I want to write a bit on the sales and distribution channels.  When Netflix started offering their streaming video service 7+years ago, I remember that I can only use the service on a browser window that runs on a Windows PC.  Fast forward to 2012, almost all of our entertainment devices are connected.  There are additional categories of products where Netflix can be watched such as televisions, blueray players, network streaming player, tablets, and smart phones.

Let’s review them briefly:

Category Dominant players Dominant Platform/OS 2012 units (source: Qualcomm Analyst Day presentation, Asymco, various) Have Netflix?
Smart phones Apple, Samsung, HTC iOS, Android 500m+ yes
Tablets Apple, Samsung iOS, Android 70m+ yes
Smart Televisions Samsung, LG, Vizio various (Linux/BSD) 30m+ yes
BluRay players Samsung, LG, Vizio various (Linux/BSD) 40m+ yes
Gaming devices, Streaming players Xbox, PS3, Roku various (Linux/BSD) 100m+ yes
Cable/satellite STB Comcast, DirectTV various (Linux/BSD) 70m+ no

From engineering stand point, Netflix has done a good job of getting their streaming software running on these various platforms relatively quickly.  From business/value chain point of view, signing these consumer product OEMs and platforms to Netflix is relatively easy given the pull of the Netflix brand from consumers.  20m+ Netflix subscribers want to watch Netflix content on their new devices and having a new TV/Blueray player that can stream Netflix will allow device manufacturers to up sell new or replacement product.  From marketing perspective, Netflix is prominently displayed in various consumer product packaging or promoted in App Stores with very good product placement.  They now have a much better distribution channels compared to other competing service.  Amazon is catching up, Verizon does not have a product, Comcast and Dish offer the service as defensive strategy to keep their subscribers.

The number of devices can mislead you into thinking that this is Netflix’s potential.  However, the real potential number for Netflix is 100m+ US households + their international expansion.  Each household will pay for 1 service and share the account with the family members.  At 20m+ subscribers, there is still room to grow.  At $8/month, this is 2-3 cups of Latte per month.  The price is so low it could almost be an impulse buy.  With much better consumer sentiment, why can’t they go to 35m-40m households with a bunch of niche contents, a few blockbusters, and original series (on-demand cable channel)? 70% of American households have shown that they are willing to pay $80/month for pay TV.

On the international side, DirectTV has proven that South America can be a big market as they grew their Latin America subscribers to 7m+ in 5 years.  Their satellite operation allowed them to have wider coverage for their service.  I am not convinced that the internet infrastructure in Latin America can reach many potential subscribers to be Netflix’s growth engine.

How much does it cost to offer pay TV service?

I want to know how much it cost for providers to offer pay TV service.  Especially, I want to compare the cost to offer pay TV services between cable TV MSOs, satellite TV providers, and over the top providers (direct over the internet).

Note I used the word subscription versus subscribers.  Comcast ended 2011 with 22.3m video subscriptions, 18.1m broadband subscriptions, and 9.3m voice subscriptions.  However, they did not mention how many unique subscribers (households) use their service.  Each household may have 1 or more subscription (i.e.: video and internet).  To get the number for Comcast, I divided Comcast’s cable business operations cost with the total number of subscriptions at 49.3m at the end of 2011.  ARPU is also lower because broadband and voice pricing are lower than cable TV.

DirectTV can generate $80/month from each of their 20m+ subscribers and spends $60/subscriber/month to attract, install, retain, and offer programming services.  About 40% of the cost goes to contents (cable channels, sports, movies, TV shows).

Netflix does not require cable/satellite dish installation, set top boxes to lease, or cable equipments to operate.  They just borrow the internet to offer the service and therefore only spends $9/subscriber/month to offer the service.  They offer the video service over the internet where the infrastructure is already built by Telcos and MSOs.  They do however spend ~40% of the revenue on streaming content.  Content Delivery Networks (CDN) cost and renting servers from Amazon is insignificant to the size of their revenue.

From technology stand point, the internet is definitely the future of pay TV as it can offer the service with the lowest operational cost.  There is no technology barrier today that prevents the same contents you get on cable channels via the internet.  However, as I mentioned in my previous blog, the cable and satellite operators who are also content owners will starve the internet-only player from contents.  At the same time, they are also offering similar services as Netflix as additional feature for their subscribers (see Xfinity Strempix, Verizon/Redbox).

The cable and satellite operators seem to be winning the war so far as they are able to maintain (lose less for Comcast) their subscribers in 2011 and come up with their own internet streaming solutions.  Netflix, on the other hand, is positioning itself as a premium cable channel where they are producing their own original programming (see NYT article).

Click the image for details

Netflix streaming video cost structure

Netflix does not distinguish between their DVD and streaming business operational cost.  All the operational cost is consolidated into 2 line items:

  • Cost of subscription – includes amortization of content/programming licensing cost, amortization of DVD libraries, postage costs to mail DVDs and CDN costs to stream content.
  • Fulfillment expenses – includes cost such as content processing facility such as shipping and receiving center for DVDs, encoding of streaming content, etc.

I just want to quickly understand the cost structure of their streaming business which includes mainly CDN and content licensing cost.  Netflix recently announced that their subscribers watched an aggregate of 2 Billion hours of content in Q4 2011.  Here is my estimate for their CDN cost based on this data.

2 Billion x 60 minutes x 60 seconds = 7,200,000,000 seconds.  Assuming an average Netflix stream of 2 Mbps, the number of bits streamed = 2 Mbps * 7.2 Trillion seconds = 1,800,000,000 GBytes (1 byte = 8 bits).

Based on data from Dan Rayburn’s streaming media blog, Netflix pays CDN providers 2 cents per GBytes of video steamed to their customers.  So their CDN cost per quarter = 1,800,000,000 GBytes * $0.02 = $36m/quarter or $144m/year.  This is only 4.5% of their 2011 revenue.

Netflix has a funny way to account for streaming contents in the financial statement.  It claimed that it needs to follow some ASC rule, who knows.  Please check for yourself on Netflix’s IR website.  This is how I interpret the streaming content cost.

For contents that are considered Asset and has balance sheet components (based on slide #10 of the content accounting guide from their IR website), I used the “Amortization of streaming content library” line item from the cash flow statement as the amount charged for content cost in 2011 ($700m).

Some content licenses are not capitalized and are not considered in the balance sheet.  I suspect that this is some kind of revenue sharing plan with the movie studios where there is a minimum amount Netflix has to pay to make the content available for streaming plus Netflix may have to pay additional fee based on the number of times the movie is streamed.  There is no easy way to get a breakdown of this since they are all included in the Cost of subscription line.

Based on slide #11 of their content accounting guide, I had to go back to Netflix’s previous year (2010) 10-K and check the contractual obligation that is due in 2011.  I took the Content obligations line that is due “Less than 1 year” as the amount that will be included in the 2011 cost of subscription ($530m).  This does not include new contracts signed in 2011 that were amortized in the immediate quarters.  It also does not include provisions for additional fee from the increasing frequency of movie streams as subscribers grow.  So this estimate is on the low side because Netflix grew their subscribers 30% in 2011 and signed a bunch of new content deals in 2011.  I am going to add 10-15% to this estimate and just round it up to $600m.

To summarize, my estimate of their streaming content cost for 2011 is $1.3B.  I did not have to work so hard to get the programming/content cost for Direct TV or Comcast because they just make it obvious in the line item.  Here is a more detailed estimate of their cost breakdown in 2011.

Click on this image to get the actual dollars

US Pay TV market overview

Recent earning reports from DirectTV, Dish, and Comcast have indicated substantial subscriber growth in the industry. I want to compare the metrics in these companies and also compare them with over the top player such as Netflix.

motion chart for US pay TV market

I have to make some assumptions to build the chart as I need to consolidate financial reports from different companies.  I will write some articles on my takeaways in the next few days.  The chart is work in progress and I will keep updating as I learn more about the companies.

Over the top streaming video primer

According to a recent study by Sandvine titled “Fall 2010 Global Internet Phenomena”, Netflix, a popular streaming video service, represents 20% of downstream internet traffic in the US during peak hours (8 – 10 PM local time).  This is the time slot where Americans typically sit back and relax to either surf the web or watch TV.  The study suggests a significant shift is happening on how Americans consume their entertainment and media content from traditional broadcast media such as cable TV to the internet where contents are on demand and available when they want and where they want.

Mass adoption of internet video does not take place over night.  As early adopters are embracing the newly found freedom of on demand internet video, there is a large percentage of people that are resisting the shift to internet video for various reasons.  Some are not comfortable with the paradigm shift of connecting their consumer electronics devices to the internet where the network setup could be daunting.  Others are not satisfied with the quality of the internet video and want the same level of experience as playing DVD/BluRay disc with instant and guaranteed high quality playback.

A number of factors affect the quality of user experience when streaming video over the internet.  Some of the problems arise from the characteristics of the underlying network used to stream video.  For example, the internet is a packet switch network that is designed for scalability but offers no guarantee on how fast packets can be delivered between two endpoints on the network.  Also packets can get lost during transmission due to network congestion, malfunctioned hardware, or routing problems.  Packet loss and retransmission during video stream has an adverse effect on the experience where video playback could glitch and stop for a period of time or image on the screen is distorted.

In addition, the farther away the video client is to the streaming server implies there are more intermediate routers along the way that need to route/forward the video packets to its destination.  The more intermediate router hops raises the time delay, packet round trip time and packet drop probability.  As round trip time increases, TCP flow control and TCP sliding window protocol further exacerbates the problem by reducing the effective bandwidth of the connection and hence degraded the video stream quality.

Another important thing to consider is video bit rates and coding algorithm.  As a rule of thumb, video quality for a particular coding algorithm gets better with higher bit rates video.  For example, standard definition DVD disc streams up to 10 Mbps MPEG2 video while 1080p high definition BluRay disc requires up to 40 Mbps.  In contrast, advanced video encoding algorithm such as H.264 can compress 1080p high definition video at 6-8 Mbps bit rates.  As a result, most streaming video services have chosen advanced algorithm such as H.264 to produce the highest quality video at the smallest bit rates.  This technique requires powerful CPU and video engine on the client devices to keep up with decoding complex algorithm and processing DRM at the same time.  In general, Moore’s law and competition in the semiconductor industry will solve this problem over time.

Other major factor to think about is the speed of video client’s broadband internet connection.   Broadband bandwitdh must exceed the video bit rates to allow continuous and uninterrupted playback of video streams.  Broadband services are available from multiple different ISPs and each ISP offers different level of service depending on the equipments that they have deployed in a particular area.   As an example, AT&T can offer broadband service up to 30 Mbps in some areas where they have installed VDSL network while they can only offer up to 2 Mbps in some rural areas.  This presents a challenge for streaming video services company to offer unique service for each customer by enabling the best quality video that their broadband speed allows.

The last factor that is often neglected is the lack of service level agreement (SLA) between subscribers and their ISP to allocate bandwidth and prioritize streaming video traffic over other traffic competing for the same broadband bandwidth.  As an example, with 6 Mbps broadband speed, bittorrent traffic from other endpoint such as a PC on the home network could potentially hog the download bandwidth and try to take all the available 6 Mbps bandwidth.  Even though this bandwidth is typically sufficient to deliver HD quality H.264 video streams, the presence of the bittorrent traffic prevents HD playback of the video.

To mitigate network congestion problem, client device can pre-buffer video streams before playback.  This allows client device to account for variation in delay and packet loss in the network and ensuring uninterrupted video stream.  Buffering does have a negative side effect because playback does not start immediately.  Typical DVD/BluRay disc plays within 20-30 seconds after inserting the disc.  Sufficient bandwidth that exceeded the streamed video bit rates is needed to meet disc playback experience.  If the bandwidth is not there, buffering the same amount of video data could take longer than 30 seconds and impacted the quality experience.

To offer uniform experience to users present at disparate locations in the world with different proximity to the streaming server, a widely adopted method is to use Content Delivery Networks (CDN) service.  CDN service provider has thousands of servers deployed worldwide typically at the ISP’s access network and these servers cache video contents locally.

Adaptive streaming technology addresses the network congestion issue and variation in broadband speeds by dynamically changing the streaming bit rates to accommodate bandwidth availability.  The server collects information about the client’s network characteristics in real time and selects an ideal video bit rates with the goal of minimizing video interruptions and providing best playback.  The server accomplishes this by storing multiple video files that have the same content but with varying video bit rates to match each client’s unique bandwidth.  Adaptive streaming not only helps in providing uninterrupted video playback but also helps in reducing operational cost as fewer bits are served through the CDN.

When other traffic on the network are competing for the same broadband bandwidth, home router gateway can offer additional qos feature to automatically prioritize and allocate more bandwidth for video streams compared to other traffic.

In summary, internet streaming video services are at the inflection point of mass market adoption where subscribers are growing exponentially and delivery experience is good enough for early adopters to create momentum for the rest of the consumers.  Streaming video experience will get even better in the future as broadband speed increases, Moore’s law raises processing power capability, and video coding algorithm improves.

Why Netflix struggles?

Netflix was gaining so much momentum that it felt like they were going to take over the world and change Hollywood forever.  Their stock soared to an all time high of $298 in the summer of 2011 valuing the company at close to $20B.  They had a stunning subscribers growth from 10m (Q2 2009) to 24m (Q2 2011).   So why has Netflix struggled recently with subscribers leaving and their stock plummeting?

Despite all the media noise putting the blame on the way management handled the price increase and the qwikster debacle, we forgot to ask the question as to why they were doing those things in the first place.  To answer this question, we should look at the cable/pay TV industry that Netflix is disrupting.

  • Comcast – represents 22m cable TV subscription with annual revenue of roughly $36B (not including NBC Universal).  Comcast also has majority ownerships of a number of cable program/channel companies (i.e.: USA, SyFy, CNBC, etc), Universal pictures, and NBC TV network and broadcasters.
  • Time Warner Cable – revenue of $20B.  Their parent company Time Warner owns movie studios who produce hit movies such as Harry Potter and other cable channels such as HBO.
  • Direct TV – revenue of $21B.  Liberty Media (owner of Starz channel and others) owns controlling stake of Direct TV.
  • Dish Network – revenue of $13B.  Dish also owns Blockbuster movie on demand subscription that are *only* available for Dish’s subscribers.

If we ignore the rest of the cable/satellite TV providers (i.e.: Cox and other smaller players) and just combine these 4 giants, this is a $90B plus industry.  A significant portion of the $90B pie goes to cable channels (Starz, USA, CNBC, etc), Broadcast TV networks (NBC, ABC, FOX, etc), and movie studios that supply the programming.  The ownership structure in this industry ensures money flows between entities that are owned by the same person or group.  This creates a powerful force in the value chain that prevents outsider or new player to enter and disrupt the industry.

Netflix offers valuable on-demand movie services at a fraction of the cost of subscribing to cable/satellite TV channel.  Netflix has premium movie contents with their Starz Channel along with older movies and TV shows for $8/month.  Their offer is so compelling that the power of references and word of mouth propelled them to insane subscriber growth for a period of time.  Consumers suddenly see a significantly lower cost alternative for their $80/month cable bill and opt for Netflix instead.

The success annoys the cable TV industry because Netflix’s gain is their loss.  But they have a powerful leverage to use against Netflix and they are using it.  Netflix still has to license contents from the cable companies, cable channels, movie studios or their subsidiaries.  As an example, Netflix has not been able to renew their deal with Starz and will lose Starz premium contents from their streaming service starting in Q1 2012.  Starz has a simple goal, they would be happy to license their contents to anybody (Netflix included) as long as they can keep the cable industry strong and allow the $90B to flow within the value chain.  It would be hard to do a deal unless Netflix sells a significant stake of the company to one of these cable companies.

There is another industry that also wants a piece of Netflix’s revenue, the internet service provider.  Netflix is distributing videos on their distribution networks for free (i.e.: cable and DSL networks require massive capital to build) .  They want a piece of Netflix’s service revenue.  Since they are not getting anything from Netflix because of FCC’s affinity for Net Neutrality, they are discussing new pricing model for broadband internet where consumers pay for the amount of bandwidth they use (as opposed to unlimited today).  The new pricing will increase the cost to stream Netflix video to the consumers.

Therefore Netflix’s recent struggle was not caused by the price changes or anything else.  The real fundamental problem is they are fighting a strong and intertwined value chain that has every reason for Netflix to fail.  This is a classic example of disruption theory at work where an entrant is trying to fit disruptive technology in an established market with powerful incumbents.  Netflix will not be able to afford premium contents going forward because the incumbents will do whatever they can to protect their business.

But Netflix will always have a niche – a powerful niche in my view.  They still have 20m+ subscriber and $2B plus to spend on contents.  They have control over their profitability on the minimum they need to spend to satisfy and grow their user base.  There will always be a subset of the consumers that want low cost video subscription service ala Netflix  ($8/month as opposed to $80/month cable bill).  They will be satisfied with the limited selection of documentaries, kids’ cartoon, teen dramas, foreign titles and be pleasantly surprised with a newer title from Dream Works once in a while.  Netflix will dominate this segment.

Tablets will also be favorable to Netflix as they are mainly media consumption devices.  A 10″ tablet device just does not feel right if you don’t have a video service such as Netflix.  Yes, Amazon has a competing service that they bundled with Amazon Prime.  But if you look at the movies and TV shows that are available at Amazon.com, you can tell that their catalog list is still mediocre because this is not their main business.  They want to hook people with their Prime membership and free 2-days shipping to generate even bigger sales for their $40B e-tail business.  The digital video subscription is just an additional marketing cost to lure people into Amazon prime.

Go to Best Buy today and look at all the new TVs and Bluray players that are displayed, *all* of them advertise the ability to stream Netflix.  This is a very large distribution channel that makes it easier for Netflix to gain new subscribers.  They still have a lot of room to gain more subscribers domestically and internationally.

It is my view that Netflix will continue to be successful as a business that caters to the low end of home entertainment market.  From hearing and reading about Reed Hastings, he just does not seem like a guy who will sell his company – unless maybe he lost control of the investors/boards.

A tale of 2 disruptions – Netflix and Youtube

The disruption theory (from Clayton Christensen) suggests that you have to find asymmetry of motivation which roughly translates to “don’t go after existing market and try to replace strong incumbents with your disruptive technology”.  He gave examples of a lot of companies from diverse industries ranging from electronic hard drives to steel manufacturing.  I want to compare 2 companies that saw the same potential disruptive power of the internet to distribute digital video in circa 2006.

YouTube chose to empower people to create their own video and easily distribute them to their friends and families via the internet.  They probably (I am guessing) did not know how to make money at the time but they knew that it was cool and people want to do it.  They ended up making money from ads on the popular videos and share revenue with the content producer.  YouTube is clearly going after completely new market with no existing incumbents.

Netflix saw the same disruption and went after the on-demand movie subscription service via the internet.   Netflix ended up competing with cable TV subscription because they are offering relatively compelling content at much lower cost.  I was one of the so called “cord cutter” back in 2009 time frame because there was a lot of value in Netflix for $10/month (streaming and DVD).  I can do the all you can eat streaming for older movies and get the new movies from DVD.  Netflix did really well for a couple years and stole a lot of subscribers away from the lucrative $90B plus cable/satellite TV business.  But the weight of the cable TV industry finally caught up and created problems for Netflix.

Some could make the argument that Netflix is offering low cost disruption.  But I argue against this because the incumbents are not fleeing to the higher margin business.  They are all fighting Netflix.  Netflix now has to explore option to re-position themselves to avoid head on battle and figure out how to leverage their 20m+ subscribers.