Tom Webster, writing and speaking

Is Twitter Becoming More of a "Representative Sample?"

Added on by Tom Webster.

eMarketer pointed us today to a recent Penn State study that examined 150,000 tweets mentioning a brand, and learned that most of these tweets were either just neutral comments or information exchanges. Only 22% of brand-naming tweets actually expressed an opinion or sentiment about the brand. The eMarketer piece notes the "good news for marketers: Twitter users were much more likely to express positive sentiments."

However, looking at the quintiles Penn State used to characterize sentiment, I'm not sure that's entirely accurate. Brand researchers generally only look at the top two boxes (and sometimes only the top box) as truly positive measures of sentiment, and in this study 33% of the tweets were labeled "Great," while 19% were "Swell." The other three categories ranged from Wretched to Bad to So-So: the latter would not generally be seen as necessarily "positive." In any case, about 52% of sentiment was positive, and 48% were "non-positive," if not necessarily negative.

On one hand, this means that brand tweets are not overwhelmingly positive, but they aren't overwhelmingly negative either. And with most brand tweets not expressing any opinion whatsoever, the whole thing looks, well, kind of "normal." All of which makes me wonder if the Twitter userbase is slowly lurching towards the center. Clearly it isn't there yet, with the majority of the noise generated on Twitter by a disproportionately small number of users, but observing that the Twittersphere appears to be at least well distributed in terms of sentiment gives social media data miners and brand managers some comfort that their products and services won't necessarily be savaged on Twitter, and in fact most brand tweets are either expressions of usage or exchanges of simple information--all of which are entirely appropriate fields in which brand managers could (and should) engage.