Griselda Carlos Arzate

Our last post looked at the two populations of Walking Dead and Vampire Diaries watchers as separate populations. The next challenge was to uniformly compare two audience groups and their social media activity on Twitter. Argus obtained buzz data for The Vampire Diaries and The Walking Dead by seeking Tweets with the words ‘vampire diaries’ or ‘walking dead’ in their body. Hence, there is a sample bias and it explains why Argus had more data for The Vampire Diaries than for The Walking Dead; while Tweets with the word ‘vampire’ were evaluated, Tweets with the word ‘zombie’ were not. Due to this disparity, it was necessary for Argus to equally compare the two groups to accurately assess buzz and viewership. The statistical tool used to enable this comparison was computing the Z-score for the two populations.

Z-scores were essential to analyzing the social dynamics of these populations. Shifts in data, due to premieres and finales, in both shows were now compared on a uniform platform without the sample bias distorting outcomes. By disregarding the sample bias, Z-scores allowed Argus to compare analysis, unlike before. A standard score, or a Z-score, specifies how many standard deviations data or an observation are below or above the mean. A positive Z-score denotes a value above the mean while a negative Z-score indicates a value below the mean. This dimension-less measurement allowed Argus to further compare episodes of The Walking Dead and The Vampire Diaries based on social media buzz. After calculating the mean and standard deviation of buzz for both series, every episode was assigned a Z-score ranging from -1 to upwards of 8, meaning the buzz for that day was, at its peak, 8 times the standard deviation above the average. This allowed Argus insights to compare the relative viewership based on the disparate samples in a meaningful way. Figure 1 is a graphical representation of the scores.

Season premieres saw the highest Z-score of any episodes. The Vampire Diaries premiere had a Z-score of 7.03 while The Walking Dead premiere had a score of 8.05, meaning that the Walking Dead premiere was more popular than the Vampire Diaries broadcasting. The mid-season and finale broadcastings also attracted higher levels of viewers given the particular theatrics associated with the episode. Interestingly enough, The Vampire Diaries finale was more popular than The Walking Dead finale, a curious fact given that Walking Dead was more popular per broadcasting for all episodes. Once the finale aired there was less viewer buzz due to the lack of new, weekly content. However, Walking Dead buzz decreased as a slower rate after its finale than Vampire Diaries, which decreased at a significantly higher rate. This could be attributed to various things, for example the level of engagement with post-season content that either series offered viewers once the season had concluded. Nevertheless, whether or not grand finales drive future levels of viewership will be of interest in the coming seasons to Argus.

Forbes recently published a list of “3 metrics that really matter [to a] startup.” Topping that list at number one is measuring how engaged users are, since users who sign up for a service but rarely use it have little value and users who constantly use the service are of great value. Similarly, Argus recently concluded that GetGlue, the social networking site for television viewers, is a better metric for measuring the relationship between social media buzz and viewership than was overall buzz data about TV shows. Argus Insights has concluded that there is a relationship between buzz on Twitter and weekly television viewership.

Two of the most popular shows on television are The Walking Dead and The Vampire Diaries. The Walking Dead is based on a comic book series and is a horror, post-apocalyptic drama and The Vampire Diaries is based on a book series and is a supernatural drama. Given each show’s fan base, Argus set out to investigate if there was a correlation between social media buzz on Twitter and viewership throughout the shows’ seasons. Although there are very few studies—including Nielsen Holding’s most recent study that relates buzz to TV ratings—about the influence of social media on television, Argus began researching viewership per episode and the buzz per day for the duration of the two television seasons. Each television series had two sets of data: viewership in millions and Twitter social media buzz in thousands. Figure 1 and Figure 2 are graphical representations of these relationships.

Figure 1. The R2 is .026, meaning there was very little correlation between viewership and social media buzz on Twitter.

Figure 2. The R2 is .2067, meaning there is a correlation between Vampire Diaries viewership and social media buzz on Twitter.

The shows experienced rises and falls of social media buzz and viewership that corresponded to weekly airings. Season premieres and finales attracted more buzz than the regular season, though the buzz generated by the finale of The Vampire Diaries paled in comparison to The Walking Dead finale. The noticeable higher levels of buzz at the beginning and ending of the season drove the creation of additional test cases as we set up the hypothesis for analysis. Argus evaluated two specific cases:

1. 3 Day Buzz: By focusing on buzz from three specific days (the day before, the day of, and the day after an airing), Argus had the opportunity to concentrate on buzz most likely associated with the actual broadcast of the episode. Buzz for those three days was totaled for a “3 Day Buzz” value (See Table 1).

2. 4 Day Buzz: Argus shifted focus to buzz from the week prior to an airing in order to assess if buzz four days before a broadcast impacted whether more people watched the next broadcast. Buzz from an actual broadcast and four days prior to the airing were totaled to attain a value for “4 Day Buzz” (See Table 2).

Table 1. Mentions of Walking Dead taken from zombie stream limited to unique users

Table 2. Mentions of Vampire Diaries taken from vampire stream limited to unique users

Season premieres and finales are unique cases because they demonstrate predictably high levels of viewership and buzz. Therefore, it is important to analyze data with and without the premieres and finales. Interestingly, R² increased after premiere and finale were removed from the data being observed. This increase is significant because it implies that the data without premiere and finale is more accurate of the fluctuations in viewers. Loyal fans are more likely to watch a series on a weekly basis than fans that are only likely to watch premieres or finales because of hype. Information about The Walking Dead indicates that while there are millions of viewers, only a fraction of those viewers use Twitter as a way to engage with the show and other fans. On the contrary, Vampire Diaries fans are more likely to use Twitter to engage with the show and other fans (see Table 2).

GetGlue is a social networking site for television fans. The application allows viewers to “check-in” while they watch live television in exchange for gaining access to exclusive stickers of that particular show. Argus obtained data about buzz that GetGlue was generating with regards to The Walking Dead and The Vampire Diaries and applied the “3 Day Buzz” and “4 Day Buzz” cases to the data with the purpose of further exploring the explanatory relationship between buzz and viewership (see Table 3 and Table 4).

Table 3. Mentions of GetGlue and Walking Dead limited to unique users

 Table 4. Mentions of GetGlue and Vampire Diaries limited to unique users

While evaluating a show without a premiere and finale was necessary to attain an accurate portrayal of viewers, GetGlue R² results present a different conclusion. While the R² values for The Walking Dead follow previous trends, The Vampire Diaries R² decrease. This outcome could be attributed to viewership demographics. Vampire Diaries viewers consist of more women who are into social viewing—which includes dual-screen viewing and using Twitter to engage with the series—that are more likely to use Getglue. Walking Dead viewers, on the other hand, consist of fewer women, are not into social viewing, and are therefore less likely to engage with GetGlue.

In conclusion, Argus found that there is a relationship between social media buzz and viewership, although it is not very robust. The lack of strong correlation can be attributed to several different factors, including demographics. Given the different themes and nature of the shows, more men watch The Walking Dead, while more women watch The Vampire Diaries (see Figure 3).

As previously mentioned, Nielsen Holding undertook a similar study. The market analysis and measurement service focused on the relationship between social media buzz and television ratings (Talbot 2011). While their focus was not viewership as was Argus’s focus, their findings shed light on the importance of social media to TV. Nielsen’s study found that a 9 percent increase in buzz in the weeks before a premiere is indicative of a 1 percent increase in ratings. In the weeks before the mid-season and finale, a 14 percent increase in buzz is indicative of a 1 percent increase in ratings (Talbot 2011). Hence, social media is directly relating to TV ratings, and possibly viewership. Nielsen also reported that television series with the greatest data included series that were watched primarily by women (Talbot 2011). Argus arrived at the same conclusion, as previously described.

Recently, Coca-Cola attempted to relate buzz with revenue gain. Patrick Coffee from PR News reported that Coca-Cola’s effort to be more active on social media has not translated into “short term revenue gains” (Coffee 2013). Coca-Cola defined buzz as “conversations taking place on social networks…[measured] by ‘counting the raw publicly available comments,’” an interesting definition to consider because they used data from sites like Facebook and Youtube while Argus only used Twitter. Coffee reports that Coca-Cola’s 60 million fans on Facebook were not enough to drive sales up. While generic buzz around The Walking Dead or The Vampire Diaries seemed to have little correlation to viewership, equivalent to Coca-Cola losing sales while having 60 million likes on Facebook, the specific type of buzz generated by GetGlue correlates much better to viewership. This demonstrates the importance of picking the right buzz metrics for analysis and ensuring your business is not pushing the wrong metric of success.

The same Coffee article reported that a poll on the Yahoo! Finance page corroborated Coca-Cola’s findings with public opinion. Users were asked if they believe “social media (Facebook, Twitter etc) [influences] what [they] buy,” to which 85% of voters, out of 500,000 responses, clicked never. While our conclusions and pubic opinion are contradictory, it will be of importance to industries around the world to re-evaluate their metrics if public opinion is indeed correct.