Why Online Buzz Shouldn't Equate to TV Ratings....

   An article in last weekend's 'New York Times' posed the opinion that 'Online Buzz Doesn't equate to Ratings', which is something I completely agree with, but for different reasons. The NYT's piece compares an index of social media conversation volumes to Nielsen viewer data to illustrate that of the Top 10 most discussed shows, only 4 are in the top 10 by US viewership. Though the comparison starts to illustrate the nuances of online conversation analysis, I think that it also touches on a relatively global & basic idea about networking & human conversation.

   As shown previously in the analysis of Twitter conversation about the Superbowl & the FA Cup Final, events are normally a safe bet to drive conversation volumes online. I use 'normally' because not every large event targets the majority of the heavy online demographic or cohesively calls to action users in a way that generates a noticeable conversation mass. However, generally we prefer to share our thoughts and experiences around a singular event with others and this leads to a relatively easy to track message/activity volume (Eurovision final, I'm shamefully looking at you).

   Extrapolating online tracking into television, I can see where the NYT article gets its premise and its contradictory tone. The Top 10 US television series are highly anticipated and, in their own way, weekly broadcasting events. Logic would dictate that such shows drive conversation for an activate audience which would generate a cohesive and noticeable mass of conversational activity through sharing the viewing event. The contradictory discrepancy in top online and viewership shows illustrates that not every top viewership show activates an on-line audience.

The lack of support for my #bumbum hashtag is listed as the primary reason that the show was canceled....

    Though the promotional purpose of combining TV viewership with online activity is to share the viewing experience and promote content, after the fact analysis is different. While large events like the Lost finale or an American Idol episode generate something to rapidly discuss (all the while interacting with a viewership that is relatively digitally savvy), I imagine less marquee events such as CSI, NCIS or a really good Law & Order (possibly a personal bias) won't galvanize action in the same volume or time frame, skewing the relationship between ratings and conversation volume.

    Alternatively, the nature of online networks means that conversation volume can spike for TV events which lack overall popularity. Due to the propagation of online viewing 'sub-cultures', its feasible to say that activity spikes for the season finale or prominent episode of a low/mid rated show could mimic the online activity of larger mainstays. Online 'sub-cultures' can drive their specific viewing tastes away from the mainstream, supporting viewing events with such interaction that they mimic the volume of conversation for large, mainstream events.

   So between the activity and content biases towards discussing certain shows online, is there a place for networks such as Twitter to predict viewership? I say there is still a large role for online messaging analysis in predicting/monitoring show performance, but not from overall message volumes. If we, as marketers & analysts, move more towards analyzing the volume of conversation in segments or over time, as opposed to relative to other shows or in general, we can begin to gather insight into content effectiveness.Even with a specific focus though, frequency analysis seems better suited for online PR analysis or advertising effectiveness, than viewership analysis.

Safe....for now...

   More effectively, the real worth of online TV content analysis comes from 'what' users are saying, not the frequency of it being said. While message frequency can begin to illustrate quick reactions to events, it doesn't posess the power to predict larger events that would equate to rating a viewing session. If we analyze what users are saying, we can begin to form a general opinion on how a user reacts to content (not just if and when) and whether they would return to share such an event again. While analysis of such content isn't currently quick or easy, it represents the true value of online conversation analysis.