Streaming tv services like Netflix, Amazon Prime Video, Lightbox and Neon are becoming more and more popular. But how to measure the popularity of content on these services? In New Zealand and overseas, television networks have long relied on Nielsen’s PeopleMeter panel to calculate ratings for tv shows. Nielsen’s local contract was recently extended to 2022, so it will continue to be the main TV ratings system here for at least several more years.
But are panels relevant in the streaming era? In my view it’s a technology destined to become obsolete, due to its small footprint. Nielsen’s PeopleMeter measures only a tiny proportion of the total tv-watching audience – somewhere between 600-900 kiwi homes at the present time.
Enter Parrot Analytics, a TV measurement service that uses software to hoover up a huge amount of online data – over a billion new data points each day, according to the company. It was created in 2013 by New Zealander Wared Seger, who saw an opportunity to disrupt old-school tv ratings companies like Nielsen.
Parrot Analytics’ algorithm is called “Demand Expressions” and it claims to measure a wide variety of data sources for online tv. This includes video streams, social media activity, photo sharing, blogging, commenting on fan and critic rating platforms, and even downloads and streams from peer-to-peer (P2P) protocols and file sharing sites.
Each type of data measured is assigned a score, similar to how Google ranks web pages for its search engines. Indeed, Parrot Analytics named its scoring system DemandRank – clearly inspired by Google’s all-powerful PageRank algorithm.
In Parrot Analytics’ case, the higher the probability that the tv show was actually viewed (and not just talked about), the higher its DemandRank score. So a stream or a download gets a higher score than a social media comment or like.
However, the elephant in the room is the absence of hard streaming data – something that media outlets almost always fail to mention when quoting Parrot Analytics figures. The two main players in the global tv streaming market, Netflix and Amazon, do not release their ratings data. So it’s likely that Parrot Analytics either makes a guesstimate of Netflix and Amazon data, or simply ignores it and uses the billion or so other data points it gathers every day.
Interestingly, Nielsen has made a very public attempt to measure Netflix ratings. The company told The New York Times last October that it is “able to determine how many viewers were streaming Netflix content through audio recognition software in the 44,000 Nielsen-rated homes across the United States.”
However, The Times pointed out that Nielsen’s ratings for Netflix “include only viewers who are using a television set.” It doesn’t measure streams via computer, tablet or smartphone. Since many teenagers do their streaming on such devices, rather than the family television set, Nielsen is probably missing a lot of data.
Netflix later poured cold water over Nielsen’s efforts to estimate its streaming data. A Netflix spokesperson told The Times that Nielsen’s numbers are “not accurate, not even close.”
Rather than try to make a public estimation of Netflix streaming figures, Parrot Analytics has chosen to use its proprietary “Demand Expressions” to rank tv shows. In its Global Television Demand Report for 2017, Parrot Analytics put Stranger Things atop its list of “top digital original series” in the United States, with an average of 58,867,391 Demand Expressions. The second-placed show, 13 Reasons Why, had 50,002,509 Demand Expressions. The report also featured charts for other countries, although disappointingly not for New Zealand.
Parrot Analytics is at pains to point out that it “does not use panel data.” Instead, it claims to measure “actual, expressed demand” for television shows. It’s a clever turn of phrase, considering that Parrot Analytics probably doesn’t have any hard streaming data. Most (if not all) of the data it uses is likely taken from social media and other online outlets. Then it’s put through Parrot Analytics’ special – yet opaque – algorithm to come up with the Demand Expressions figure.
Don’t get me wrong: I applaud Parrot Analytics for coming up with an algorithm-based solution to how to measure streaming ratings. There’s no question it’s better than panel data, since Nielsen cannot accurately measure streaming data from its panels. But a word of caution: we can’t yet declare that Parrot Analytics has solved tv ratings for the streaming era. Only Netflix, Amazon and other streaming platforms know for sure how many people view their shows.
That said, using social media and other online data points was a master stroke by Parrot Analytics. Regardless of whether it’s as good as streaming data (hint: it’s not), it’s still undeniably useful data for tv production companies.
What’s more, if recent comments by Parrot Analytics CEO Wared Seger are to be believed, the data it collects will soon be used to not only measure streaming tv – but to predict the success of future tv shows.
Seger wrote last week that “we can now look at the popularity of certain genres, sub-genres, themes, cast members, show-runners and even more accurately predict what kinds of content will resonate with viewers in certain territories around the world.”
That’s an admirable goal, but it does beg the question: will this result in more formulaic and less original tv shows? Conservatism and lack of imagination are already common attributes in the movie industry – as evidenced by the endless amount of re-makes, sequels and Marvel movies we’re subjected to every year. By contrast, the tv industry over the past couple of decades has been vibrant and groundbreaking. It’s a big reason binging became popular on streaming services, as we all rushed to consume compelling, original shows like Stranger Things and House of Cards.
Despite these concerns, and my reservations about the opaqueness of its algorithm, I have no doubt that Parrot Analytics has brought tv ratings into the 21st century. Now, if only it can persuade Netflix, Amazon and other streaming companies to let it access some of that valuable streaming data.