Tuesday, October 15, 2013

Twitter TV Ratings Are Here, But No One Knows What They Really Mean




Twitter’s offices in San Francisco. Photo: Ariel Zambelich/WIRED



Nielsen recently released its first Twitter TV Ratings top 10 list, a new metric that measures not how many people watch a given television show — the basis for traditional TV ratings — but how rather how much activity they generate on Twitter. Just how different are the two metrics? Take look at the Breaking Bad finale: While it came in No. 1 on the list as the most tweeted about show in America, its traditional TV ratings didn’t even crack the top 10.


Nor did many of the most-watched shows make the same sort of splash on Twitter. The most-watched shows during the the week Breaking Bad topped the Twitter Ratings chart were Sunday Night FootballThe Big Bang TheoryNCISNCIS: Los AngelesThe Crazy Ones, two other NFL programs and The Voice. How many of them showed up on the Twitter TV top 10? Only The Voice.


It turns out the shows most seen on Twitter aren’t always the shows most seen on TV, and the discrepancy illustrates just how much more there is to learn about what a tweet about a TV show means. It also offers networks and advertisers a way to gauge the relationship between viewership and Twitter traffic, but knowing the volume of tweets is just the beginning, and there’s a lot of data in those tweets that’s still silent.


For example, were people tweeting so much about Breaking Bad simply because it was the finale? Do shows like Glee and Jimmy Kimmel Live do well because they have younger audiences that are more tuned-in to Twitter? Are people tweeting that they love The Voice or that they’re angry at who’s winning? That information is out there, but it doesn’t come out in numbers on the reach of a given tweet (Nielsen tracks the number of tweets, people tweeting about a show, and also how many people see those tweets)


“Social provides a lot of opportunity, but it’s difficult to relate what’s being said online – especially on Twitter – to what’s being done in the real world. People say one thing and they do another,” Brian Blau, a Gartner analyst who researches social analytics tools, told WIRED. “How do you help these businesses figure out what the differences are?”


In other words, we may know that a lot of people are tweeting about a show, but are they watching it? And, if so, are they enjoying it? Who these tweeters are and how they feel about, say, an episode of Scandal isn’t something you can determine from knowing that there are nearly 713,000 tweets about it.



One thing that has potential to change this, Blau noted, is Twitter’s new deal with Comcast, which was announced last Wednesday. The two companies are pairing up on a new feature called “See It,” which will essentially turn Twitter into a remote control. Launching in November, “See It” will create a Twitter card for any show mentioned in a tweet that will allow users to click to watch, On Demand stream, or DVR the show mentioned. That feature, Blau said, will give much more effective data on how often a tweet leads to someone tuning in. And that seems to be what Comcast, which has 24 million U.S. subscribers, is looking for. As the company’s chief business development officer Sam Schwartz told All Things D, “we want to make the conversation on Twitter lead to consumption.”




Nielsen has determined there is a two-way relationship between Twitter and ratings, even if no one is quite sure exactly how it works. Back in August, the media measurement company released a study that looked at 221 primetime episodes and found that the live ratings for a given episode had a “statistically significant” impact on tweets related to the show for nearly half of the episodes sampled and – conversely – the volume of tweets caused a significant change to the live ratings of 29 percent of the episodes. Translation: For nearly a third of the episodes sampled, the more tweets, the higher the ratings. Interesting, but even Nielsen is still a little unsure what it means.


“This round of causation research was only looking to see if there’s a ‘there there’ with respect to tweets influencing ratings,” a Nielsen rep told Variety when the report was released. “Now that we’ve seen statistically significant evidence of this, the next wave of research will be around understanding how/why.”


Granted, most of this information is generated in order to give networks and advertisers a better idea of the kind of traction TV shows are getting on social networks, but the metrics are likely going to be interesting to fans as well as ad buyers. Think about it. Joss Whedon’s cult favorite space cowboy show Firefly never got great Nielsen numbers, but if there had been Twitter TV Ratings back when it was on the air, fans could’ve pointed to the show’s presumably significant level of social engagement as a sign of its niche popularity.


So what’s next in terms of looking at engagement beyond the numbers? Nielsen’s SocialGuide, which the company acquired last year to work on its social TV measurements, is working to broaden its Twitter metrics to include demographic data like age and gender in 2014, according to a Nielsen spokesperson. There’s also a plan to include tweets in Spanish.


What Nielsen probably won’t track, simply because it doesn’t lend itself to straight-forward numbers, is the idea of sentiment — how people actually feel about the things they are discussing. Is someone tweeting about the Breaking Bad finale because they loved it or hated it? There no distinction in the current Twitter TV Ratings. And while several analytics firms – including Bluefin Labs, which Twitter acquired earlier this year – do monitor sentiment, it’s a notoriously difficult thing to measure. There’s no foolproof way for text-analyzing machines to understand sarcasm and slang, for example.


“The vendors that are experts at this claim 90 percent of the conversation that they measure, they have an accurate measure of sentiment or an accurate measure of the subject of the conversation,” Blau said. “Skeptics tell me that that’s very high – 90 percent is at the upper bound – but the realistic numbers are much lower, in the 50 to 70 percent [range]. Will it ever be 100 percent? The answer is no.”


Of course, the bigger question is: Does it matter? Some say “any press is good press,” and perhaps all tweets are good tweets. Even if everyone’s talking about how terrible or WTF a show is, if the volume is high enough it might cause people to tune in just to see what all the fuss is about. It’ll be especially interesting to see how this plays out in the new arrangement between Twitter and Comcast. Based on the screenshot examples released by the cable provider, it appears the Twitter cards will at least initially be attached to tweets from the official show Twitter feeds, but if/when the cards are attached to random users tweets, you might see an update that reads “Tonight’s #GreysAnatomy was the worst ever!” followed by a card asking if you’d like to watch it now.


In a blog post announcing the gambit, Comcast Cable’s head of business development Sam Schwartz used the example of seeing friends tweeting about Sharknado, which got a lot of tweets even if many of them were mocking (or at least ironic). “If I had only seen an ad about a sharks-meet-a-natural-disaster movie, frankly, there would be little chance that I would tune in,” Schwartz said. “However, all these tweets pique my curiosity, I click on the See It button in one of the tweets, and then use it to set a reminder to watch the movie later that night.” So maybe love it or hate it or love-hate it — it really doesn’t matter just so long as there are eyeballs.


Metrics like age, gender and sentiment are still only the beginning. There’s still a lot of other data — and combination of data — embedded in the world of Twitter for networks and advertisers to mine, even if analyzing it may prove the bigger challenge. “It’s complicated,” said Clark Fredricksen, vice president of research firm eMarketer told WIRED. “On the one hand, you have the internet, which is the most accountable, measurable media channel in history, compared to TV, which is arguably the most difficult channel to measure.” In other words, social media provides an embarrassment of riches that viewership numbers don’t. We just need to figure out how to read it.


“It’s probably fair to say we’re past version 1.0 of social analytics tools. Version 1.0 is giving people basic high-level statistics – counts, followers, and maybe the first order of statistics saying what does some of this data mean?” Blau added. “[Now] we’re into version 2.0 of social analytics, where they’ve realized how to get access to the data, they’ve had their first level, they’ve realized what their customers want now in terms of information and all those smart data scientists in the world are hopefully coming up with algorithms to give them that.”



Source: http://feeds.wired.com/c/35185/f/661370/s/327b5726/sc/15/l/0L0Swired0N0Cunderwire0C20A130C10A0Ctwitter0Etv0Eratings0C/story01.htm
Category: walking dead   thursday night football   Electric Zoo   amber heard   nelson mandela  

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.