Sean Casey talks about Social Content Ratings, the most comprehensive measurement of TV-related social media activity available, which released last month.
As media fragmentation increases, it is more and more difficult for networks to generate the levels of viewership they have historically been used to. Driving engagement and cultivating loyalty is correspondingly more challenging, too. The net result is a threat to the value networks can deliver to their advertisers, which is, of course, a threat to their own success.
This is not a new problem. But it has become significantly more intense over the last five years, with the growing sophistication of devices other than TV on which to watch video, and with the arrival of new types of publishers that did not exist as recently as five years ago—most particularly, the streaming video-on-demand players such as Amazon, Hulu and Netflix.
But what technology takes with one hand, it often gives back with the other. And one of the things that technology has made possible is an explosion in communication and sharing through social media that can help networks engage with audiences in unprecedented and beneficial ways.
For a while now, networks have been using “Social TV,” people’s responses to TV on social media, both during live broadcast and at other times, in order to increase tune in, and to engender loyalty to and engagement with their programs. They understand social TV is an increasingly important part of their viewers’ world: in 2015, 60% of consumers used their smartphones or tablets while watching TV at least once per week, many of them using their devices to be social about the programming they watch. That same year, over 24 million unique users sent more than 800 million TV-related Tweets, generating 80 billion impressions.
But monetizing this activity is not so simple. To do so, networks need to understand, on a cross-platform basis, the effectiveness of both their “owned content” (the social content they themselves are pushing out on social platforms) and the “organic content” being generated by their TV audiences in generating viewership, loyalty, and engagement.
Historically, they have been able understand, by platform, how effective their owned content is in generating engagement with their audiences—because they have the data necessary from the social platforms. They know how many engagements their content is generating, and they can make content decisions accordingly.
However, understanding the organic content their programs generate has been a problem. Here, they have had to rely on third-party providers for measurement, a reliance that brought with it two problems.
First, today’s still-very-young social content ecosystem has a half-functioning, half-chaotic Wild West feel to it, with social content measured in many different ways – just as, when “TV” turned into “video” as the platforms and devices on which it could be viewed proliferated, all sorts of companies grew up to measure digital’s effectiveness by a dozen different metrics— “impressions,” “clickthroughs,” and so on.
Second, third party measurement providers’ access to the organic data is limited. Any third party, by using Facebook’s public APIs, can get access to the truly public social TV content on Facebook. But most mentions of TV content on Facebook happen one-on-one, or in smaller group exchanges between friends. This represents the overwhelming majority of TV mentions on Facebook. What third parties can generally measure is merely the tip of the iceberg. What is under the waterline is every bit as important.
The first problem is being solved: In an attempt to bring some order to the social content measurement universe, the Media Rating Council (MRC) has recommended a consistent framework by which social content should be understood. Specifically, the MRC recommends that social content should be analyzed in terms of owned and organic measurement across authorship, engagement and “reach”, with demos. That is, it should be measured in terms of authorship (who posted the original content), engagement (comments, likes, retweets) and reach (who sees the content), each categorized as to whether it is driven by owned or organic content—all broken down by demographic. This framework, to which we at Nielsen are adhering, addresses the ‘Wild West’ element of the ecosystem today,
But the second problem—the volume of Facebook content below the waterline, as it were—persists for measurers in general. Having entered into a partnership with Facebook, we ourselves have access to this content. As such, Nielsen is the only comprehensive measurer of social TV content, and we measure it in line with the MRC framework.
Nielsen receives data directly from Facebook in order to measure total social TV activity in an aggregated way so that it is totally anonymous and therefore fully respectful of users’ privacy. Both public social media activity and activity between friends is measured. (Facebook Messenger activity is exluded under this arrangement, and so is not included in our measurements.)
Without this comprehensive measurement, networks cannot fully understand when, where and how social audiences are engaging with their social content and their programming—which leaves them flying largely blind in determining the level of investment they should be making in the various social platforms on which this organic conversation is taking place. We are the first to bring to the market comprehensive measurement of Social TV of the kind now available on a syndicated basis for TV—or rather, for video, whether viewed on live TV, time-shifted TV, streamed, on traditional TV sets, smartphones or tablets.
It may be that, one day, others will have access to enough video-related social media to provide alternative services. We have constructed our Social Content Ratings to maintain an advantage even under that scenario.
In order to measure social content, you have to develop “classifiers” —essentially, words or phrases – that can be used as a kind of filter when analyzing the billions of messages on social platforms in order to identify and capture those related to a particular program. What is not well known is that the social activity while a program is being broadcast live is quite different from when it is being shown at other times. This means you need two sets of classifiers—one for the linear (i.e., live) broadcast, and one for the rest of the week.
This is not a trivial distinction. In the week of Jan.18, 2016, 48,000 Tweets were sent about one particular cable reality program. When a program is airing live, we use a much larger set of classifiers about both the program and its talent. In this case, we used 70 classifiers from three hours before through three hours after the live broadcast (“the linear window”). The Tweets captured by this linear classifier set during this time accounted for 61% of Tweets for the program during that week. We then used a narrower set of classifiers directly related to the program—13 terms—for round-the-clock measurement conducted through the week. There are many different episodes of this program airing in repeat as well. The remaining 39% of Tweets concerned other episodes as well as other general activity related to this program.
What happens if you simply use the narrower set of terms for all social measurement of a program? Our analysis determined that 37% of live program activity, on average, is identified by linear classifiers used only during live broadcast. If you don’t have a broad set of classifiers relevant to live broadcast but not to airing at other times, you are going to miss almost 40% of the conversation.
Why can’t any company create a special, larger set of classifiers for live broadcast? Well, they can and they can’t. To establish a set of linear classifiers for any given program is no big deal: you just need to know when the program is on, and study the social activity around it. That’s all well and good. But to measure social media activity on a minute-by-minute basis for hundreds of TV channels and thousands of programs aired in the U.S, only an industrial-strength measurement infrastructure and an experienced staff of content editors will do the trick—along with licensing of a trusted, universal, electronic programming guide with descriptions of each episode, and information about the characters. Few companies can make the investment necessary to measure this activity robustly at scale. That is what we have done.
It’s early days: Social TV will not come of age as a commercial phenomenon until there is an agreed standard of measurement that all measurement companies are delivering against. The introduction of the MRC’s social media guidelines—particularly the recommendation that the core measurement focus on authorship, engagement, and reach across owned and organic activity—will, we believe, be the standard that will allow the market to maximize the value of social TV. Until then, we will be in a situation similar to the early days of measurement of TV on smartphones and tablets vs. TV—different measures, hard to compare, and hard to combine into a comprehensive view of what is going on. Measuring social TV comprehensively, measuring it differently for each program at time of broadcast than in general, and measuring it comparably, will all be fundamental to unleashing its power. We are ready today.
Sean Casey is president of Nielsen Social.