We ’re thrilled to drink back The Entertainment Strategy Guy who’ll give a comprehensive breakdown of Netflix’s change in how they give criteria to us, the general public including advantages and disadvantages to the new system A couple of months back at Vox Media’s Code Conference, Netflixco-CEO Ted Sarandos released this little map which set the internet a fire That’s right, for the first time, Netflix revealed not just how numerous people were watching a piece of content, but also how long. At the time, I wondered why they were being so generous with their data.
And I ’m remonstrating myself for not realizing the factual answer Netflix had prepared their daily report and decided to switch criteria. Preliminarily, they blazoned how numerous subscribers watched two twinkles of a given Television show or film in the first 28 days. (I called the “ datecdotes”.) Going forward, they ’ll release the total hours viewed for Television series and flicks by week. In their words
IMAGE 2 Netflix Earnings Quotation Given how important I had preliminarily laidby Netflix’s precious datecdotes — consolidated on this website into this excellent resource — this Twitter exchange added up my passions at the time Still, all those datecdotes I collected over the times would be for aught, If Netflix changed criteria. Basically empty, since I could n’t compare them to other data points.
Netflix said they were planning to be “ more transparent” going forward. Well, I do n’t suppose anyone realized just how transparent that would be! Two weeks agone, Netflix set up a brand new website to partake the total hours viewed for the top 10 shows and flicks across both English andnon-English content. Forty data points each week! And they handed 20 weeks of data at the time!
Hallelujah, that’s a lot of data! And while I ’m thankful for it, let’s not de-emphasize the change. Netflix did change criteria, which means going forward it’ll be much harder to compare performance from 2020 and before to 2021 and further. It’s also worth pondering why and what the different criteria can tell us. Overall, different criteria are n’t really good or bad, they just are. Hours viewed is a veritably useful metric, but so is unique guests ( homes) watching. Let’s explain.
What’s More Total Hours Viewed or Unique Homes?
.When it comes to data, I ’mgreedy.However, my usual response is to ask for further data, If you asked me what piece of data I ’d like. Data judges worship the demon Mammon when it comes to data Further is better. Then’s a real- life print of me clinging my data
IMAGE 3 Larfleeze
This is n’t just rapacity, however. Only having one metric is the problem. To truly understand trends, you need multiple measures of success Say you want to know the stylish three point shooter in basketball. Firing chance is a useful place to start. Then’s a histogram of all the NBA players who ’ve tried a three- point shot this season
IMAGE 4 NBA Histogram
Of course, some guys on this list slightly shoot threes. Which is why you have the two tails of some folks who have ridiculously high probabilities and some who are at zero. To get a better sense for who’s a good three- point shooter, let’s add in how numerous shots players have taken
.
IMAGE 5 NBA Scatterplot
That’s a much more accurate description of three- point shot quality. And not unexpectedly, the stylish three- point shooter is Steph Curry Or take an issue like seminaries. Test scores matters a lot, but so do scalerates.However, seminaries could protest out low performing scholars, If you just measured testscores.However, seminaries would pass kiddies who ca n’t read, If you just measured scale rates. So you need to track both criteria, scale rates and test scores. One metric hardly ever captures the reality of the situation.
Still, pennants would release both unique guests and streaming hours for all of their content, If I had my druthers. But I do n’t rule the world, so most pennants do n’t release any data. And Netflix just changed which metric they released. So let’s bandy both unique guests and total hours viewed to see which is better.
The Advantages of Unique Guests (or Homes, Subscribers, Members or Accounts) Up until now, Netflix has released a “ uniques” metric. Meaning they count all the accounts that watched at least two twinkles of a given show or movie. Netflix has used a many different words to describe this — guests, accounts, subscribers, homes, etc — but all get to the idea that they ’re measuring a unique log-in who watched some length of a piece of content.
( Presumably different biographies are n’t counted multiple times, again because of the homes language.) Of course, indeed a straightforward standard like this requires some private analysis. For illustration, how long does a unique account need to watch to “ count”? This was the subject of Netflix’s first metric shift. Back in 2020, after releasing viewership figures grounded on subscribers who watched for “ 70” of a piece of content, Netflix shifted to subscribers who watched “ at least 2 twinkles” of a piece of content. Then’s a breakdown of how numerous we got of each type
IMAGE 6 Netflix Datecdotes by Type Reportedly, Netflix calls 70 observers “ watchers”, guests who watch 2 twinkles “ starters” and folks who watch 90 “ finishers”. Really any of these criteria are useful, depending what you ’remeasuring.However, either 70 or 90 is presumably stylish, If you want the stylish idea of guests who actually are watching a piece of content.
Unique observers is enough useful because it gives a good approximation of “ reach” or the number of folks engaging with a piece of content. It’s also veritably useful for the marketing folks, because, presumably, if they can get a lot of people to start commodity, they did theirjob.However, also it means the content was n’t good, (If they do n’t finish. Unique observers also puts shows on roughly equal footing. A lower show like Lupin can contend with a giganticmulti-season series like Stranger Effects because it is n’t about total volume of occurrences, but just unique subscribers watching it.
Unique guests, however, does n’t track the number of folks who started a show but failed to finish, commodity hours viewed can compare. And for some series, completion rates really do matter. For illustration, when Netflix released Part Two of season one of Lupin only 54 million homes watched two twinkles, down 41 from Part One released in January. Then’s a map I made for subscribe growth or decay after Netflix’s last earnings call
IMAGE 7 Lupin Decay Lupin, in particular, was short. Only 4 occurrences in Part 1 and Part 2. Still, we would n’t know how many folks finished it, If Netflix had released all eight occurrences at formerly. We ’d have assumed that utmost of those original 76 million watched the whole thing. Broken up, however, we know a lot of people did n’t finish it.
The Advantages of Total Hours Viewed Total hours get us much closer to understanding true completion rate. In this case, you can take the total figures of hours, peak by the number of occurrences available, with a presumed completion rate, and guess how numerous homes watched Guessed” being the operative word. For film, since it’s a one-off event, with a known time period, this can be fairly precise. But for Television it’s a disaster. When I say “ presumed completion rate”, those figures can veritably hectically. So the error bars are fairly broad. Still, it’s presumably more useful than just subscribers.
If there’s one reason this change will be “ stupendous sauce” — a specialized term — it’s that total hours viewed is how Nielsen measuresU.S. viewership. Meaning we can compareU.S. viewership to the rest of the world in an “ apples-to-apples” manner The hours viewed metric does have its own impulses. In particular, it favors content that’s rewatched a lot. Specifically, kiddies content. It’s not like knockouts of millions of Americans are watching CoComelon each week. Rather, a veritably small number of kiddies watch it all the time. Unique homes helps reveal that difference.
(By the way, if you ’d like another explainer on the difference between colorful criteria, check out this composition I wrote from a many times back.) Still, take this success in content is identified, If you take nothing down from this composition Think of big megahit. Like Squid Game. It’s the top Netflix series in terms of total guests watching, total hours viewed, and, presumably, completion rate. And it’s largely rated on IMDb. In other words, successful shows are successful in utmost criteria.
And I can illustrate this with a visual
That’s a smatter plot of unique guests versus hours per guests for both Television and film Basically, the thing for any show is to get as high to the up and right as possible. That means lots of people are watching and watching for a long time. The trend line goes up and to the right, because the quantum of total hours viewed is identified with the number of unique guests.
(I made the two maps by syncing all the Netflix datecdotes over the last three times with the top ten most popular hours maps they ’ve released. In some cases, I used estimates from the Netflix Hours Viewed for series with four weeks of viewership. So do I wish Netflix would release unique subscribers for all their shows going forward too? Obviously. Further data is better.
But I ca n’t be frenetic. Netflix dropped 800 new data points. The most from a banderole ever. And while I ’ll miss the old “ datecdotes”, by this time coming time we ’ll have a lot of data to dissect Still, check out my other jottings over at my ( recently redesigned) website, subscribe up for my newsletter, (If you enjoyed this data dive In particular, each week I publish a “ Streaming Conditions Report” collecting multiple data sources to unload how content is performing on streaming each week. Check it out.)
