(This is the third part of a multi-part series exploring one specific Netflix number. To read the other pieces:
Okay, enough skepticism. If we take the latest datecdote from Netflix, at face value, what can we learn from it?
Well, to start, let’s take a look at the history of “80 million” in Netflix releases…
What other evidence of “suspicious” numbers do you have?
In a future article, I’ll write about another “theme” of this website called, “Theme X: Be Skeptical”. Especially with competitors. Don’t give them the benefit of the doubt!
One of the corollaries to that theme relates to data. The corollary is, “Be wary of large, rounded numbers.” Data isn’t often rounded so evenly. This applies to scientific studies, political causes and other social phenomena. Oh, and entertainment success stories.
I’d add, if the same company keeps repeating the same big number that would be weird, right?
Netflix has this problem with their original movies, and I don’t think anyone has pointed this out yet. Researching the “80 million customer accounts” I naturally googled to try to find every news source uncritically repeating this datecdote. Imagine my surprise when the first occurrence of 80 million accounts wasn’t October in the Q3 shareholder report, but actually in June!
See, their “Summer of Love” romantic comedies weren’t the first time they had “80 million customers” watch something. For example…
– In June, Dana Feldman on Forbes also reported that 80 million customers had watched a “romance film” on the service. Rereading it, this looks like it includes both originals and licensed films. This came from a Netflix tweet.
Clearly there was some rounding going on. And for press releases or on information provided on background, Netflix is under no obligation to be precise. But let’s assume the numbers are close enough. If the baseline assumption is all Netflix movies combined get 80 million customers accounts involved, the “Summer of Love” films didn’t really boost viewership that much did they? Either for romance films or original films. Eighty million customers is just what big groups of movies promoted by Netflix tend to get.
But if I wanted to be skeptical, I mean, what are the odds that exactly 80 million customers watched an original in from January to June, which is the same number that over the previous year watched a “romance movie” and then, after two more months of Netflix rom-coms being released in the “Summer of Love”, they had all in 80 million customers watch an original romantic comedy? Is that crazy overlap, or three part coincidence?
What if we take all the recurring 80 million customers at face value? What can we learn from this number?
We can triangulate the floor for Netflix “Monthly Active Users”.
This is the biggest way that streaming video distributors, social platforms and subscriptions services in general try to game the narrative. A customer or user includes anyone who “samples” a subscription. So you order from Blue Apron, or start a Hulu free trial, or sign up for SnapChat. Those are all users or customers.
But Monthly Active Users (MAUs) is a much better approximation of who is actually using your service regularly. (Or weekly or daily active users, which are even smaller time periods.) That means actual people you can monetize through ads or monthly billing. With Netflix, we have no clue what their monthly or weekly active users are. Most social platforms include this in their SEC filings. Netflix does not—it isn’t a social platform—and instead focuses on “subscribers”.
That doesn’t mean “Monthly Active Users” wouldn’t be incredibly valuable information. Take my viewing behavior this summer. I did use the service in July, but not June or August. Because I wasn’t using the service monthly, we cancelled in September. So I would have counted in the subscriber count, but not in their MAUs for two of the three months.
My working hypothesis is that if Netflix raises prices, the most likely people to drop out are those not using the service. (Yeah, not the most original theory.) Having the Netflix MAU number on a monthly basis going backwards would be a key way to understand how price increases would likely impact customer dropout rates. If it is low—say 40 to 50% of total subscribers are monthly active users—a price increase would likely motivate a lot of customer to drop out. If it is high, say 80% or more, they probably wouldn’t. High MAUs would also be a sign that their content is doing well to keep people involved.
One of the benefits of the Netflix “Summer of Love” 80 million customer accounts number they officially released is that we now have an idea what this number is. The number has to come from somewhere, which means people actually using the platform. Knowing that at least 80 million accounts interacted with the platform over a 3 to 4 month period (remember, Netflix was vague on the time period), this gives us a good idea of the floor for Netflix’s MAUs.
The fact that the near 80 million number was the same for all original films, summer of love films and romance films in general over 6, 4 and 12 month periods make me think the 80 million number is probably fairly reliable.
So 80 million divided by 137 million worldwide customers gets to 58% of subscribers use the service monthly or quarterly, at least.
Is that good or bad? Well, I don’t know. It’s still a datecdote, so we can’t compare it to Amazon Prime/Video/Studios or Hulu or Youtube. Would this lead to a high drop out rate or low? Or how much higher would it go if there are people not captured by one of the three film categories above? I don’t know, but at least we know that the floor is 58%!
What else can we learn about Netflix content?
Well, I’ve had the hypothesis that “binge released shows” have really high drop off rates after they are released. (Fine, it’s more than a theory. I saw it with my own eyes.)
Both MDb and Google Trends data show is that Netflix movies tend to support this theory with Netflix originals. Essentially, two weeks after they launch, all the buzz is gone. This is the opposite of Crazy Rich Asians, A Star is Born and even Venom, who maintained interest for weeks after they launched. Even The Christmas Chronicles looks set to buck this trend. (On the TV side, Stranger Things definitely bucked this trend.)
If you drop all the episodes on one day, and they aren’t a monster hit like House of Cards or Stranger Things, then you lose any chance to maintain interest. This is why “windowing” is so powerful and was/is so important to movie releases. Crazy Rich Asians is still generating buzz and awareness as it releases on home entertainment and VOD.
With Netflix rom-coms, after that first weekend, they fade away. Does that prove their model is superior to traditional releases? I don’t think so, but a lot of smart analysts disagree with me on that.
Since we’re trying to be fair to Netflix, talk about that data you left out last week, social media. Didn’t you see that Instagram chart?
I did. So let’s address it.
First, the good news for Netflix is that their original movies really do kill it on Instagram, specifically. You probably saw the bar chart showing the before and after rises of their various stars that was also in their Q3 Letter to Shareholders. But sometimes bar charts are very visually appealing. So here is a comparison of their “Summer of Love” stars to Crazy Rich Asians Stars.
So if—and it’s a huge if—social media is directly related to viewership, then game over Crazy Rich Asians. I mean, nearly every star in the Netflix movies had a multiple of 9 or so more Instagram followers more than Henry Golding, Constance Wu and Awkwafina.
Of course, I looked up two more additional social media networks. The conclusion, initially, is that Instagram is just clearly more popular than Twitter and Facebook if these few stars are representative examples. That said, the gap between the other social platforms isn’t nearly as large, or in some cases smaller, on the other platforms than Instagram.
What do we make of this? Well, the knowledge that social media is not—repeat NOT—a representative sample of America. It skews much, much younger and demographically each platform has different skews. This is really important when doing data analysis. If your population is not representative, then it isn’t representative. It’s not a bad thing, but it just means we have to be cautious.
That’s why the only data set I think is ubiquitous is Google since it has something like 99% penetration of the American market. The other social platforms just don’t. Oh, and on all social platforms you can always buy followers, which is another giant red flag. If you can just pay to boost a ranking, it isn’t really reliable. You just never know if followers are organic or juiced.
So yes, this is a sign that maybe these two films had more viewership than Crazy Rich Asians—and more data that The Kissing Booth and To All the Boys I’ve Loved Before were the only two two successful Netflix rom-coms—but it isn’t the whole story. It is the data point that gives me the most caution when anointing Crazy Rich Asians the winner.