Yesterday, I covered the strategy thinking behind Netflix’s latest data drop. Today, I’m going to dig into the initial takeaways I have from the data and how I plan to use it. And then I’ll be finished talking about this new data release.
Kidding!
I’m far from done. I still need to analyze the data while adding all the metadata I do for new content—which will generate a ton of ideas and visuals—and also explain the value between these two different metrics, “hours viewed vs customers”. So expect more on this over the next few weeks.
(Due to Thanksgiving holiday/festivities and Nielsen ratings coming out next Monday, the weekly streaming ratings report will move to early next week. As always, sign up for my newsletter to get all my columns, streaming ratings reports, and articles in your inbox.)
Netflix’s Big Data Drop…Quickly Explained
Here were my emotions when I saw the news that Netflix was releasing forty data points each week. And had provided 20 weeks worth of data:
At first, I was worried.
Then, I was overwhelmed.
And then I got excited.
I was worried at first because Netflix’s lack of transparency is kind of my business model right now. A commonly repeated, but wildly incorrect assertion, is that “No one knows what is doing well” in TV ratings. This is just flat wrong, as my “Steaming Ratings Report” has shown since March. However, even though we know more than ever, that doesn’t mean it isn’t messy. Wading through all the data and putting it in context requires some work.
So I was briefly worried this would hurt my product. But as I thought about it, it won’t. First, while it’s great to have global data, country-level data still provides crucial context. Long time readers have heard me say that the streaming wars will be fought country-by-country. To extend the analogy, we won’t know who is winning each battlefield without country-specific ratings. Netflix’s new global data is very useful, but other data is useful too. I’ll keep analyzing the U.S. data in particular.
Second, even with all this data, you’ll need a ton of context and analysis to understand it. Which is partly why Netflix is releasing it; they can overwhelm with volume, and still rely on generally positive headlines from the trades. I’m convinced I can provide deeper analysis than most, even with this new data. Especially when it comes to “dogs not barking”, the Netflix Originals that fail to chart upon release. (Trust me there are plenty of those.)
And third, even if Netflix is being forthcoming with their data, their peers are not. So I still have a job!
(And if that sounds partially like a sales pitch for why you should pay to subscribe to a future newsletter offering, well yeah. I’ve been thinking about this. Which means a lot of thinking on differentiating my offering from other subscription products.)
This is a Lot of Data To Analyze
After being very briefly worried, I was then overwhelmed. I mean, 800 new data points isn’t a lot from a “big data” perspective, but it’s a lot for me to ingest at one time. Before I analyze a film or TV show in a data set, I collect a lot of information about it, some of it automated and some manually. Take genre. I don’t trust most online sources for genre. They’re too broad and unwieldy. Since this list is so foreign title heavy, that’s going to take a bit of work to incorporate.
But it will be worth it, as long as you give me a pinch to digest it. Of the 800 new entries, by my initial accounting, there are 299 unique entries, either series or films. (Many TV series have multiple seasons and this new list counts each season separately.) Of those, I only had 107 (or 36%) in my U.S.-focused content database. So that’s 190 new entries I need to update and categorize.
That’s a lot of data to absorb at once. And foreign titles can be tough to find quality information on. But I’m working on it.
Let’s Get Excited For This New Data
Once I got over the amount of data, I got excited. For example, we can start to make comparisons like this. First, here’s the Nielsen ratings for the week of 18-Oct, then the Netflix “Top 25ish” chart for the same week:
What do you think of the new table? Send me a note or hit me up on Twitter. Like life itself, good visualizations evolve over time so we’ll figure out how to best display this new information, and this is my first draft.
Can we start to pull out some takeaways? Cautiously yes.
As one would expect, Netflix Originals do very well in this measurement, especially compared to Nielsen’s measurements. Partly this is because Netflix pushes Originals very hard, but partly it is a function of availability. For many series, Netflix doesn’t have global rights. It will be much harder for a show only available in America, for example, to make the global charts.
We can also see the general strength of English-language programming, but not absolute dominance by any means. Here’s a table showing the total hours viewed by each category:
Another thing I like about my “top 25ish” ranks is they show you where we can stop “knowing” what was the most watched. Specifically, after Locke & Key season 1, we don’t know if there was another English language TV series that was more watched than Carinha de Anjo season 1. So the rankings are “unknown” after that point.
Still, if you did rank the list in order you can see that it’s still a hit’s driven business:
Again, we don’t know for sure these are the top 40 films or TV series on Netflix globally. In fact they probably aren’t. But still the trend in viewership is that top series and films aren’t a bit better than their peers, they’re multiples better, and this gives a flavor of that. And I’d stop making this point, but I keep reading articles that insist this isn’t the case!
Here’s another point I’ve been making about the U.S. ratings that applies on the global level too. Due to release cycles and the decay of customers over time, I’ve noted that many TV series follow the “binge release curve”. Friend of the website Emily Horgan pulled some kids data, and you can see that curve now on the global scale:
Are you excited yet?
Because after just the previous analyses we’ve done—moved to the global scale—we can start to uncover other relationship. For example, we have Nielsen-tracked U.S. viewership in hours and now Netflix global viewership in hours. Let’s just combine them:
Ta da! And thus we see that Maid is the most U.S.-centric piece of content, with 31% of its total viewership coming from the U.S. In comparison, The Five Juanas did the worst, since it didn’t even rank in the U.S. (The best it could have done in the U.S. is 5.1 million, which My Name did.)
Insights!
And that’s only working with a partial data set.
One More Emotion: Relief
I’ll admit: when Netflix said they’d be more forthcoming with data, I never thought they’d be this forthcoming. Forty data points each week? With numbers attached? That’s a lot of data.
So yeah, I’m bummed we won’t get future unique subscriber counts and we won’t get past “hours viewed” for titles they haven’t released yet. But Netflix will provide a good chunk of data on a consistent schedule, and I’m glad to have it.
