Tom Webster, writing and speaking

Descriptive vs. Predictive

Added on by Tom Webster.

statistics.jpgA few days ago I saw a lot of people tweeting about this post, which amalgamates some recent data about Facebook and Twitter into an infographic comparing the two services. What many of these tweets focused on (and, indeed, so did the article accompanying the infographic) was the higher percentage of Twitter users who are more likely to purchase products and services from brands they follow than their counterparts on Facebook. Yes, it does appear that Twitter users are more likely to say that they follow brands, and more likely to say that they do indeed purchase from those brands, than are Facebook users. Of course, Twitter users are Facebook users. These aren't different people per se; Twitter users merely represent a small, potentially more active subset of Facebook users, and behaviors are different. The danger with data like these is not that it isn't factually correct - indeed, our own research on Twitter verifies this increased likelihood to follow brands.

No, the danger with slavishly retweeting this sort of data is believing that it is predictive, rather than descriptive. That Twitter users have a higher predilection for discussing and following brands may or may not be endemic to Twitter itself, but it is certainly descriptive of the current user base of Twitter, which may or may not be a moving target. In no way should one assume that this data means Twitter is inherently a better place than Facebook for brands to be - especially given the relative sizes of their user bases. Rather, it says more about the fact that Facebook more closely mirrors mainstream populations, which - let's face it - don't spend all that much time talking about brands.

My only point here is that one shouldn't look at this data as data about Twitter. It's data about the types of people whom Twitter currently aggregates, who are (as I mentioned) a subset of Facebook users anyway. When we look at statistics as prescriptive, especially when they relate to a moving target, we run the risk of missing the forest. Statistics are descriptive. Any given snapshot of this data is interesting tweet fodder, but the trend is your friend - it's how these data move over time that provides the genuine insight here. The real question is not whether or not Twitter is fertile ground for brands and marketers - it's whether or not it will be. The smart money continues to look for information, not evidence.