When it comes to TV program engagement, networks and advertisers see value in social media. One reason for that is because high Twitter activity around TV programs is indicative of high viewer engagement. Networks also turn to social media to understand how program-related buzz relates to audience tune-in, and advertisers and agencies have seen that paid media placements in highly social programs can significantly boost earned media for their brands.
But can we identify what makes a program social?
To find out, Nielsen started with the inherent characteristics of television programming, such as what type of network a program airs on and the average size of the live TV audience, setting out to investigate how networks, agencies and advertisers could leverage that information to anticipate how social a program will be, as well as identify whether a given program is under- or over-performing on Twitter.
To do this, Nielsen looked at 457 English- and Spanish-language prime-time series programs, measuring average Twitter activity and TV audiences for new episodes that aired from September 2014 to January 2015. Nielsen then created a model to estimate the average number of Tweets per episode for a program using eight program characteristics as variables.
At the onset of the analysis, Nielsen determined that it considered all of the characteristics it analyzed to be underlying program traits that networks and content producers could not easily change. As such, the traits did not include more movable levers, such as promotions activities or decisions on program content and quality.
All eight of the tested variables proved to be statistically significant. In other words, all eight had a relationship to the average volume of program-related Tweets sent each week for any given program.
In fact, this integrated model explained 67% of the variance in Twitter activity. Said another way, basic differences between programs, such as TV audience size and what type of series each program is, can explain the majority of the difference in Twitter volumes between programs. As one might expect, audience size alone (i.e., live audiences age 12-34 and 2-99) plays a large role in telling the level of Twitter activity around a program, explaining 51% of variance in average Tweets per episode.
This model can help gauge whether the amount of social activity that a show is generating is higher or lower than expected, given the basic traits of that program (for instance, how social should a one-hour broadcast drama that averages a live TV audience of 3 million people each week be?).
There are three ways that TV industry players can benefit from analyzing program characteristics to anticipate levels of social activity:
It’s notable that basic program traits can explain such a large percentage (two-thirds) of variance in Twitter activity. That also suggests that these basic traits do not explain all variance in performance. Other factors, such as content quality, promotion strategies or to what extent the cast is active on social media during airtime, could explain the remaining one-third of variance in performance. These remaining variables are levers that networks and content producers could potentially use to drive social activity.
Overall, these finding introduce opportunities to more holistically understand social TV behavior, while also setting up the potential for future testing to evaluate what other factors contribute to social success for certain programs and networks.