A tale of two Twitters

There are at least two different Twitters in action at one and the same time. I’ll get to what I mean in a moment. For now, these two randomly selected, real, live tweets from my feed will hoist a hint at where we’re headed:

• @AccentureAnalytics Companies that apply #bigdata and #analytics to their operations show 5-6 per cent higher profitability
• @MileyCyrus wigged woting wis wack wust wayin [sic]

First, Twitter. It has mindshare and global reach, but it’s not nearly as 'mass market' a platform as Facebook. To drop some science on you, according to the Pew Research Center, the percent of online US adults who used Facebook last year was 71 per cent -- Twitter hovers at about 18 per cent, one tick higher than Instagram. Most of these users aren’t very active. Twitter has about 50 million people who visit monthly, while Facebook has almost four times as many.

So we’ve got fewer than one in five plugged-in adults in the US checking out Twitter for realsies. They’re more mobile than Facebook users -- crazy mobile, in fact, with 70 per cent of Twitter minutes spent on a mobile device compared to less than 50 per cent of Facebook minutes last year, according to comScore.

Other than being on the run (from the law?), what else do we know about Twitterers? This chatty 20 per cent of us who have inserted #hashtags, short URLs and @MileyCyrus into the collective unconscious?

The truth may surprise you. Recently, a company called CivicScience, based in Pittsburgh and stocked with Carnegie Mellon brainsters, released a study called “An Analysis of Twitter Users vs. Non-Users”. (You can download it and other good stuff here.) CivicScience is an innovative research firm, relying on brief surveys embedded in their website and a network of outside publishers, combined with a cookie-based user store of previous answers. They develop interesting correlations about consumer feelings toward celebrities, media, brands, and so on. (Example: Fans of Ellen DeGeneres are more likely to eat at IHOP.)

The Twitter analysis was based on answers from a large population sample (about 170,000), which mirrored the US population in terms of penetration. (That is, about 20 per cent of the sample used Twitter.) CivicScience then compared how the two groups answered various demographic and attitudinal questions to gain insight into how Twitter users differ from the non-users.

Some of the findings were simple head nodders. Twitter users are slightly more female and somewhat younger than non-users. They are more urban, more likely to be camping out in mom’s basement, less likely to have little tweets (ie children). They tend to 'showroom' while shopping, consult online reviews, think they’re trendy, play fantasy sports, and avoid Applebee’s.

For me, the more interesting findings were these:

• Twitter users are more likely to fall into the lowest or highest ends of the household income spectrum
• Twitter users are less likely to vote -- although more likely to say they are engaged in politics
• 71 per cent of Twitters users proudly boast they are 'addicted' to digital devices
• Twitterers watch more TV than the general population

The study wasn’t intended to segment Twitter users into groups, which is why I’m admittedly reading into it here. But these numbers reminded me of a previous job, doing social analytics for large consumer brands, when I had the task of poring through brand-related tweets until my brain felt like it had been pecked to death by chickens. In that job, for all the tweets we pulled about our brands -- ranging from a high-end car company to a low-end financial services factory -- consistently 60 to 70 per cent of them were categorised as “irrelevant”. That meant they were content-free, indecipherable or odd. Examples might include:

• @Tweeter 1  Saw that loco [BrandX] on the street today boy
• @Tweeter2  @MileyCyrus makes me think of a [BrandX] in a tank top

It’s not that such tweets are not about [BrandX]. It’s just that their information value is either neutral or impossible to figure. On the other hand, some of the most knowledgeable people in the world -- particularly about cars, by the way -- are on Twitter. In my experience, the ratio was about 80/20: that is, 80 per cent pop-cult gibberish and posturing and 20 per cent actual thought.

Before you jump up and down, let me assure you I’m not assuming that irrelevant tweets come from low-income tweeters, or that the low voting-to-engagement ratio means tweeters all lie. The data does not go there. I am just saying that it’s my impression Twitter -- moreso than other social channels -- is bi-model, a split-personality, a dual universe.

So, here’s the hypothesis: I think there are two different Twitters.

1. Smart Twitter (20 per cent) -- populated by experts opining to experts, news media, bloggers, and brands and consultants churning out content marketing; most of these people are directly or indirectly trying to sell something
2. Other Twitter (80 per cent)

What do you think?

PS If you’d like to see an example of Twitter #1 (and sometimes Twitter #2) in glorious action, please follow me on (yes) Twitter @martykihn.

Martin Kihn is a research director at Gartner Research. This post was first published on Gartner’s blog platform. Republished with permission.