As everyone likely knows, Netflix changed metrics on us. Previously, the world’s biggest streamer released the number of subscribers (or households/accounts) that watched 2 minutes of a given show or film in the first 28 days. (I called these datecdotes.) Now, and going forward, they’re releasing the total hours viewed for all their content.
I’ve written about this a bunch: a most important story here, and a quick explainer here.
But I’m not finished! I published a guest article at Whats-On-Netflix explaining the difference between hours viewed and unique customers. Mainly to answer a question I’ve seen floating out there: what’s the better metric, hours or customers? So I broke down the pros and cons of each metric. (The actual answer? Both metrics is my preference!) Also, in the past I wrote an explainer on various metrics here.
And today, you get another visual. Specifically, since Netflix provided top ten lists for both subscribers and hours, we can compare the two to see if hours viewed is correlated with subscribers. And here are the results:
The moral is that while I do wish I had both hours and subscribers, since the two metrics are highly correlated, having only “total hours viewed” will be fine. (And for additional context, seriously check out that guest article at Whats-On-Netflix.)
Other Thoughts and Insights
How did I make these scatter plots?
Since I’ve been hoarding the “2 minute subscriber datecdotes” for years, I simply checked to see which of those datecdotes also had their total hours revealed, either through the top ten most popular list Netflix started releasing or in the new total hours viewed per week data set.
For example, Bird Box and Extraction are both included in Netflix’s current “most popular English-language film” list, and previously had 2 minute datecdotes, so I could just compare their old 2 minute viewed counts to the total hours viewed. (Both metrics are in the first 28 days.) Sixteen films and sixteen TV shows fit into this category.
For a few shows and films, though, Netflix provided their datecdotes in the last earnings report (Vivo and The Guilty, for example, at 46M and 69M subscribers) so I used the total hours and took the first three to four weeks of viewership (which means an estimated 94 and 106 million hours). Now, four weeks of data isn’t a full 28 days—most had Friday release dates, meaning it’s 23 days of data—but it’s close enough. There were five films and two TV shows in these categories.
I also includes some films with only 3 weeks of viewership data like The Kissing Booth 3 and He’s All That. If anything, these missing data points would likely put the shows closer to the trend line, showing an even tighter connection.
What are the weirdest differences/omissions between the total hours viewed and datecdotes?
On the TV side,, both seasons of 13 Reasons Why ranking in the top ten for total hours viewed but Netflix never providing 2 minute datecdotes is strange. It shows heavy rewatching, presumably, and also why Netflix kept renewing it.
Tiger King season 1 and The Queen’s Gambit are the biggest TV series by subscriber count (64 and 62 million respectively) to not make the TV total hours viewed lists.
On the film side, Project Power and Fatherhood are the two biggest film datecdotes (75 and 74 million subscribers respectively) without total hours viewed numbers. Red Notice is the obviously biggest missing datecdote. In other words, it’s the first true victim of the post-datecdotes era.
Where did the trend line come from? What does it mean?
Since the scatterplot looks correlated, I simply drew a linear trend line through it and made it intersect at zero. And voila, we see that for every million people who watch, you can expect about 2.3 million hours viewed for a film. In other words, a little more than the length of a two hour movie. Cool!
We can use this to guess at unique subscribers who watched.
For example, we know that through 3 weeks of release Red Notice has been watched for 328 million hours, now Netflix’s biggest title. So if you divide by 2.3, you’d estimate that means 140 million unique subscribers have watched, or about 65% of Netflix’s 213 million current subscribers. Could someone confirm this for us? Hey, here’s The Rock:
I also received confirmation (these metrics take time to be verified from the Wall Street folks) that well over 50% (closer to 60%) of all @netflix accounts have viewed RED NOTICE with an INCREDIBLE 92% AUDIENCE SCORE 🍿🍿🤯 THANK U GUYS!! @Chloe_Melas 👊🏾 https://t.co/QQ3ygdVbUt
— Dwayne Johnson (@TheRock) December 1, 2021
So the estimate is a little high—he puts it below 60%—but still it gets us in the ballpark. Like Bird Box, given its high demand, it likely has more customers per account watching it. (I doubt it is heavily rewatched.)
I Do Like How Netflix Provided 20 Weeks of Data.
The most valuable part of this data set is the sheer quantity, and that it will be updated going forward. To debunk my least favorite quote, there aren’t three types of lies, lies, damned lies and statistics. There are lies, damned lies and statistics without context.
For example, go way back to Netflix’s first datecdote, Bird Box. They told us only that Bird Box was a huge hit with 80 million unique subscribers watching. But was it? Now, three years and 130 data points later, we know it was huge at the time. And it’s still their second biggest film to date. And it came out way back in 2018.
Right now, coming on six months of a data and the four “most popular lists”, we have fairly good context of what works. Though, by next July, we’ll have very, very good context.
My Last Regret
Lastly, while I’m so excited to have all this Netflix data, it really does suck that all those datecdotes from the past are, basically, worthless now. Unfortunately, by shifting to hours viewed, any insight into Netflix’s viewing patterns before July 2021 a locked in stone.
We have our 130 data points about the era before July 2021, and we’ll never know more.
