11 Comments

Summary:

Netflix doesn’t release any ratings for its original content – but it’s keeping a very close watch on who is viewing what in order to greenlight and promote shows.

IMAG0839

If you’ve been on Netflix lately, you may have seen a trailer for House of Cards, the company’s first big original content play. But chances are, the version of the trailer you got to see is different from the one that the site presented to me. That’s because Netflix is using ten different cuts, geared towards different audiences – and presents them to viewers based on their past viewing behaviour.

That kind of targeting is part of Netflix’s approach to use “a balance of intuition and analytics” for its original content, explained Netflix Chief Content Officer Ted Sarandos during an interview on the sidelines of the at All Things Digital’s Dive into Media conference in Dana Point, California Tuesday. Netflix is using “really big data,” as Sarandos put it, to evaluate everything from promotion of its original content to the question of which shows it is going to pick up next.

And we are talking not just simply audience size levels, but very fine-grained data, down to the engagement level with each individual piece of content, as well as the mix of content each subscriber is watching. “When you say ten million people watch a show, that really doesn’t tell you anything,” he said.

As a consequence, Netflix isn’t talking about how many people have been watching House of Cards ever since the show went up on the service earlier this month. “We are not doing ratings,” Sarandos said on stage Tuesday afternoon, adding the vague assesment: “We are thrilled with the numbers.” Asked what he would tell if House of Cards star Kevin Spacey called to ask about those numbers, Sarandos replied: “I don’t tell him.”

Creatives, however, don’t seem  upset about such secrecy. Arrested Development creator Mitchell Hurwitz and star Will Arnett, who joined Sarandos on stage, pointed to the problems they had with the industry’s love for ratings when their show was on air. Back then, Nielsen wasn’t including catch-up views from DVRs and online platforms in their ratings, and many of the show’s viewers simply didn’t register. “Appliance stores knew we were huge, but Nielsen didn’t give a sh*t about it,” Hurwitz joked. “The Nielsen system I think is incredibly flawed,” agreed Arnett.

That’s why a ratings-free approach is so appealing to creators, and potentially so disruptive to the industry as a whole. “The model is going to change,” predicted Hurwitz. Creatives would have to work with a different economic model, because a deal like the one with Netflix doesn’t include residuals. “The trade-off is that we are encouraged to make a more interesting show as opposed to flattening it out,” he said. Arnett agreed: “It’s very inviting to people in the creative community to have a place like Netflix,” he said.

Sarandos echoed that sentiment when talking to me, admitting that his phone has been pretty much ringing non-stop ever since the company announced that it got David Fincher to do House of Cards. “We became a first stop as opposed to a last stop,” after that announcement, he said.

This story was corrected on Sunday 2/17 with the correct first name of the creator of Arrested Development. His name is Mitchell Hurwitz, not Michael.

  1. Does anyone else find that Netflix’s clever algorithms are too smart for their own good? I find browsing their selection a painful experience because they never seem to show me the same list of films twice. On several occasions I have been unable to find films that I previously stumbled across and planned on watching later simply because the list that showed them was no longer visible to me. Unless you remember enough of a films name to use the brute force search tool you are scuppered.

    Share
    1. I agree. I also find their search very limiting. The web search is so so but try searching on your Roku or other TV connected device and it’s a pain. It’s makes it very grueling when you’re trying to find something to watch besides the recommendations they are showing. After a while you want to just give up.

      Share
    2. I find their clever algorithms not so clever. You know what, I don’t like Foreign Action movies, or British Crime Shows, or Violent SF, I like good movies and TV shows. Just because I watched Spooks doesn’t mean I want to watch Matlock.

      Share
  2. i wonder how netflix is calculating the viewership

    Share
  3. Watching the series now off my recordings via PlayLater. Recorded one right after another, but I’m watching as I have time off of Netflix’s clock. As far as they know, I binged on the entire series. Plus, my wife and I are watching at different times. So much for big data!

    Share
  4. This is a great example of a good use of big data – a very specific outcome based on very specific variables. Too many “big data” products try and start with neither of those thing and I remain skeptical about how useful they are generically.

    Share
  5. The writer of the show is not Michael Hurwitz, it’s Mitchell Hurwitz.
    http://www.imdb.com/name/nm0403804/

    Share
    1. Thanks for pointing this it, it had been corrected.

      Share
      1. thanks

        Share
  6. First, it should be noted that big data analysis is the new normal. The article seems to imply that only Netflix is using big data analytics to help create new learnings… It is hard to believe Nielsen is not utilizing data scientists, adaptive and predictive models, big data technologies and new forms of data to measure and predict consumer behavior. Yet, what is not clear in this article is, does Netflix share its data? I mean if Nielsen has been one of the TV industry’s standards to provide unbiased research, wouldn’t Netflix want to play with unbiased research organizations?

    Netflix to me is like HBO + Verizon. It benefits from vertical integration that other competitors do not. Hence they have a supreme data set for their own programming analytics. I think it is hard for the TV industry to understand the importance of ratings without platforms like Netflix and other majors like Hulu, XBOX Live in its data sets.

    From a TV network’s decision making perspective, ratings alone do not give a full story on what a show brings to a network. Yet, it could be one from a portfolio of measurements used to analyze how a show is performing? I agree that the usage of ratings may be ineffective in isolation. Their could be a new measurement or a suite of measurements that Nielsen uses that is more about measuring engagement + sentiment on TV brands instead of ‘shows’. Yet, one would need to identify the purpose / outcome of that suite of measurements.

    Share
  7. Jonelle Nonete Wednesday, March 27, 2013

    First, it should be noted that big data analysis is the new normal. The article seems to imply that only Netflix is using big data analytics to help create new learnings… It is hard to believe Nielsen is not utilizing data scientists, adaptive and predictive models, big data technologies and new forms of data to measure and predict consumer behavior. Yet, what is not clear in this article is, does Netflix share its data? I mean if Nielsen has been one of the TV industry’s standards to provide unbiased research, wouldn’t Netflix want to play with unbiased research organizations?

    Netflix to me is like HBO + Verizon. It benefits from vertical integration that other competitors do not. Hence they have a supreme data set for their own programming analytics. I think it is hard for the TV industry to understand the importance of ratings without platforms like Netflix and other majors like Hulu, XBOX Live in its data sets.

    From a TV network’s decision making perspective, ratings alone do not give a full story on what a show brings to a network. Yet, it could be one from a portfolio of measurements used to analyze how a show is performing? I agree that the usage of ratings may be ineffective in isolation. Their could be a new measurement or a suite of measurements that Nielsen uses that is more about measuring engagement + sentiment on TV brands instead of ‘shows’. Yet, one would need to identify the purpose / outcome of that suite of measurements.

    Share

Comments have been disabled for this post