PeekAnalytics: A New Way to Measure Twitter Influence
Despite all the money thrown at it, measuring social media influence has been a tenuous match of art, science and the secret algorithm sauce of each monitoring platform.
That may be about to change, and not a moment too soon for agencies and brands who are very likely to be basing their budgets on incomplete, and often bogus audience and influence measurements.
PeekYou's beta of PeekAnalytics Social Audience Report comes closer than any of the more than 50 social media monitoring platforms and tools I've tested to providing, and explaining, meaningful audience analysis of Twitter followers that brands can confidently use to create their budgets.
While you can get a river of data, a lot of pretty charts and graphs, with prices ranging from free to thousands of dollars a month, it's hard to find a platform that actually checks that your audience consists of real people whose identity can be verified, or that will explain how it really arrived at its conclusions.
PeekAnalytics' closed, invite-only beta currently measures the digital footprint of Twitter followers across 60 social sites and millions of blogs to provide accurate, actionable, data-driven Twitter insights, including, age, social membership, interests and much more about their audience.
To demonstrate how Peek Analytics works, Peek compared my digital footprint to those of Mitt Romney and Herman Cain. Turns out, despite having 10x fewer followers, I have almost as much pull as both of them. (No, thanks, I'm not going to run for office.)
Seeking accurate digital footprints
PeekAnalytics "Pull" measures passive and active Audience Quality which is fundamentally different from Klout, PeerIndex and Twitalyzer's engagement metrics. To gauge audience quality, PeekAnalytics provides over 30 metrics, including name, income, industry, location, social graph, social activity and public content. Twitalyzer, Klout etc. simply can't produce those metrics.
Klout's nebulous scoring measures Twitter comments, retweets, @replies and mentions, among other things, some of which are little more than nonsense, as Paui Gillin recently pointed out.
As PeekYou founder Mike Hussey told me, actual influence and reach is so hard to determine because a celebrity with a million followers may actually have a high percentage of bots, spammers, inactive and bought accounts – all of which sharply reduces the actual value of his/her audience. PeekAnalytics claims its algorithm can eliminate the vast majority of useless accounts.
Peek goal: moving beyond the secret sauce
"When it came to us creating a score," Hussey told me, "I knew we had to put forward a number that means something – so if Coke had a Pull of 700x and PepsiCo had a Pull of 350x, those numbers both mean something independently (350x means 350x more potential audience than the average social media user) and in comparison (A Pull of 700x vs. 350x would have twice as much influence or Pull on the social web because of the quality of the respective audiences)."
The goal, he says - beyond numbers, and secret sauce – is to analyze followers' whole digital footprint in a way brands can understand and act on.
What matters most, says Josh Mackey, PeekYou general manager of business and product development, beyond simply understanding the make up of your social audience is how well-connected they are on the Internet, how far and wide their words can spread there, and how much weight they can throw around—compared with the average person.
Mackey says the calculation of "average is based on two assumptions: "The average person has 135 consumer followers on Twitter and 444 connections across the social graph (from the sixty sites that PeekYou indexes). So the average person is two degrees away from 444^2 or 197,136 connections."
The consumers that Peek analyses are verified, individuals, who can be matched back to a phonebook, public records, or matched across multiple trusted web sources. Peek defines Consumer Ratio as the percentage of an audience who are verifiable as Consumers.
Any accounts that can't be fully identified because there is insufficient information available across the accounts' available digital footprints are categorized as "unidentifiable." Those are taken out of the count because they are considered to be a mixed bag of anonymous accounts, bots, fun accounts, and some spam accounts.
"As you know," says Mackey, "Klout, even before the update, was attacked because the scores would fluctuate up and down for seemingly no reason. With Pull, unless you do something that causes your audience to mass 'unfollow' you or even create a large following your account; your score will be much more consistent. With Pull, you will be able to understand why it is going up, (e.g. you spoke at a conference)."
Comparison reports: me vs. Romney, Cain
The comparison reports on my audience vs. Romney and Cain's are below. Here's a summary of what they show:
- Romney and Cain have very similar audiences, and they actually share 42,000 followers.
- Even though I have 1/10th their audience, I have more than a third of the "Pull" than either of them, meaning the people who've chosen to follow me are much more influential.
- My audience is more highly educated and makes more money. So if we were products, my audience would be more valuable to marketers.
- I have much higher percentage of identified people than either of them.
- From the people Peek Analytics could 'identify', my audience is much more transparent, connected and active on social media.
PeekAnalytics, still in closed beta, is currently rolling out to select agency partners.