Commentary

Does Twitter Predict TV Success?

How predictive of media success is social media chatter anyway? The film industry has been monitoring buzz for years, of course, and there have always been debates about how much Friday audience response to a new film on social media can impact Saturday performance. Some correlation between the chatter on these channels and at least the initial popularity of a media launch would seem obvious. But is it quantifiable and rigorously predictive?

Nielsen, in its never-ending pursuit of extending the value of its Twitter TV service, suggests that Twitter talk can be used in part (and with a lot of caveats) as a reliable data point for TV buying and programming adjustments even before a premiere airs. The company scrutinized Fall 2014 premieres on both broadcast and cable. First it looked only at the impact of raw promotion: how many impressions were served on TV about a program to the 18-34 segment. Not surprisingly, there was a high correlation between the number of promotion impressions served and the size of the premiere+7 audience.

advertisement

advertisement

To see if adding Twitter TV data could predict premiere audience even better than promotional exposure, Nielsen created a model that tested three variables: promotional activity, Twitter activity and whether the show was on network or cable.

Nielsen says that its model was in fact a better predictor of initial audience size than modeling from TV promotion alone.  “The model explains 65% of the variance in the premiere audience sizes, compared to 48% using promotions alone,” the post at Nielsen claims. The company believe networks and advertisers could have used the model to identify the top and bottom 10 premiere performers last season.

On one level this seems common-sensical if you compare the different data inputs. For promotions, the number of  impressions seen was measured -- not if they registered or were liked. Promotions are dependent on a show's core concept and content to have a lasting impact. Twitter TV does more closely register impact, and specifically how the show’s promotions are resonating.

Nielsen is quick to warn that these measurements do not suggest that Twitter activity causes bigger audiences. Still, media marketers would like to think that they can tweak this channel to stimulate buzz that translates into larger audiences. It still comes down to something more than quantification: Hot how much people are talking about you, but what exactly they are saying.

None of this gets us beyond the first episode of a show anyway. If a show sucks, the network just gets a bigger audience to know how bad (or more likely, how bland) it is and not come back. At that point, programmers probably could and should use a more qualitative than quantitative approach to social media response to gauge where the show hits or misses and with whom, and to think harder about how to salvage things.

Of course the unintended result of TV programmers and buyers relying too much on predictive analytics is that it could abort, too early, as many good shows as bad. Some of the most enduring series in TV history were given a chance to grow both their artistry and their audience over time despite initial ratings. One can imagine a cascading effect of advertisers pulling their support too soon on projects, leading the networks themselves to pull the plug earlier than they do already. Maybe someday, fall TV promotions will just be a series of market tests against which programmers could use predictive modeling on social responses to decide whether to even premiere a show.

After all, just a decade ago, who would have expected a show like Netflix’s "House of Cards" to come from the service running the numbers on such a granular understanding of its audience’s media behaviors?     

Next story loading loading..