Before you read all the think pieces that are about to come out discussing Netflix's latest data dump, you should probably get some context first

Here's the press statement and Excel table that has been generating all the excitement.

Since launching our weekly Top 10 and Most Popular lists in 2021, Netflix has provided more information about what people are watching than any other streamer except YouTube. And now we believe it’s time to go further. 

Starting today we will publish What We Watched: A Netflix Engagement Report twice a year. This is a comprehensive report of what people watched on Netflix over a six month period¹, including: 

• Hours viewed for every title — original and licensed — watched for over 50,000 hours²; 

• The premiere date³ for any Netflix TV series or film; and 

• Whether a title was available globally. 

In total, this report covers more than 18,000 titles — representing 99% of all viewing on Netflix — and nearly 100 billion hours viewed.

If you haven't been following streaming closely there are a few subtleties here you are likely to miss.

From the get-go Netflix is characteristically spinning this to the best possible advantage. While it is possible that the famously opaque and sometimes deceptive company has suddenly decided that openness is the best policy, this decision probably has a lot to do with the fact that they and all the other streamers are about to be forced to release far more viewership data than they have in the past due to recent contracts with the writers and actors guilds, and have decided to get out in front things and spin them as much as possible to their advantage.

To be fair, this is a major improvement in transparency and, for those seriously following the industry, very useful data. However, television and particularly streaming is a complicated business and a naive reading of these numbers, which you will certainly get from a lot of the upcoming think pieces, is likely to be misleading for a number of reasons.

From a subscription service business standpoint, total series viewing hours is not the most important or even the second most important metric. While there is a relationship between what you watch and what convinces you to sign up then not cancel the next month, it is complicated and sometimes counterintuitive.

For more than 70 years, television has always been a medium more about what we are willing to watch rather than what we want to watch. These numbers tell us a lot about the willing, not so much about the want.

Personal example. About a year ago, I decided I really wanted to revisit Justified so I signed up for Hulu (I grew up in the Ozarks, culturally not that far from Harlan County and while I didn't know people like the characters in the show, I knew people who knew those people). I ran through all the episodes fairly quickly and enjoyed it immensely, then looked through the streamer's other options and killed some time for a couple of months before dropping the service. If you were just going by my viewing numbers, the Elmore Leonard adaptation would be at or near the top but it wouldn't particularly jump out at you. I was using a "might as well see what they've got" criterion since I wasn't planning to stick around that long. I watched anything I thought I might like so I logged quite a few hours

From an enrollment and churn standpoint, however, Justified was the only show that really mattered. Everything else I watched at best kept me on the service for one additional month, which when you figure in the costs of those shows, was very much a losing proposition for Hulu.

Netflix has always been secretive and sometimes misleading about the data it releases, particularly regarding the drivers of members signing up and dropping off. A few years ago, the company even went so far as to claim that it didn't even look at churn and therefore couldn't be expected to release detailed numbers to investors, a claim which everyone I know who works in this or related industries insist had to be a lie.

If these numbers don't tell us much about the drivers of the industry, don't they at least tell us something about ourselves as a society based on what we choose to watch?

Not as much as you might think.

For starters, we have spiky data combining shows with very different viewing distributions in a way that is guaranteed to produce unrepresentative snapshots. Perhaps not coincidentally, these distortions happen to serve the narrative that Netflix and the other streamers have been pushing.

Anytime your window covers a period of time less than a year, the top 100 shows will be dominated by new, heavily promoted releases while older shows will show up considerably lower. If, however, you take the long view and look at two or three years worth of data, the picture will shift. Shows like NCIS or the Gilmore Girls may not break the top 100 in any given short period, but they will turn in consistently strong numbers year after year while many, probably most, of the hot, buzzy shows will see their numbers sharply as soon as their moment in the sun passes.

This is analogous to a discovery that Barnes & Noble made many years ago, back in the pre-Amazon times, where they realized that titles like Great Expectations or Huckleberry Finn were never among their top sellers for any given week but the titles that did hit the top of the charts often ended up in the remainder bins. That was why the company went into the classics publishing business and had a long and highly probable run with it that to a lesser extent continues to this day.

One interesting example here is the viewing hours for the last season of the Crown. This is one of the best known and most hyped and critically lauded shows that Netflix has ever run and it has been presented as a huge hit. According to the Excel table, the official release of the final season was in the second week of November, 2022. We don't know what the numbers look like that year, but we do know the total hours that the show got between January and June of this year. It was in 152nd place. the next best rated season was at 388. While we can't say for sure, it's reasonable to assume that if the time frame had been shifted a couple of months, the season would have broken the top ten or at least the top twenty.

For a more extreme case, three years ago Tiger King was on of the biggest hits the service had seen. In the first six months of this year, the first and second seasons ranked 4,133 and 6,632 respectively in viewing hours.

Authors of think pieces on this data should also remember that there are some very heavy thumbs on these scales. Netflix has spent literally billions on marketing and PR, almost all of it going to a relatively small number of shows that were seen as potential hits, or were potential sources of hype and awards. Disney has spent even more over that same time period. 

And if you still believe that your streaming service's recommendation algorithm is driven strictly by your preferences and viewing habits, rather than whatever the service is trying to push at the moment, you are a sweet and trusting person and I hope you never change. 

When you zoom back enough to see past the spikes and take into account the inflation factor of marketing and PR, it appears that people still like to watch AMC dramas and CBS police procedurals.