Twitter Influence Graders: Behind the Green Curtain

Mike Johansson
Mike Johansson Senior Lecturer at Rochester Institute of Technology and principal at Fixitology, RIT and Fixitology

Posted on February 28th 2010

Does influence on Twitter matter? And, if it does, what is the best way to measure it?

On the first question it seems reasonable to say influence only matters if all or part of the reason you are on Twitter is for business … then “influence” matters.

So if it matters to you, how do you measure influence?

In Twitter 1.0 most believed: Followers = Influence. Clearly the world of Twitter has moved on.

Today there are many tools that claim to measure an individual's influence using up to 140 characters.

I decided to look at the three Twitter influence graders I use and ask questions of the people behind the tools. Here's what I found:

Twitter Grader

Twitter Grader is just one of the many free analytics tools from the always helpful folks at Hubspot. I heard back from Dharmesh Shah, the founder at HubSpot himself (and
@dharmesh on Twitter).

Twitter Grader ranks Tweeps on a scale up to 100 and allows comparisons to others by city and state, as well as analysis of followers and those you are following.

Dharmesh says Twitter Grader looks at a number of factors but one of the most important is the degree of "engagement" for a user.

“So, if a given user seems to be getting more retweets and responses to their tweets, they would get a higher grade.

“This makes sense, because this ‘engagement' score acts as a decent proxy for influence,” he says. “We also look at the number of followers — with some adjustments for when an account has a lot of low-quality followers.”

Dharmesh went on to say that what people should know is that Twitter Grader measures on a curve (i.e. a relative scale). “So, as the Twitter user-base evolves, the scores adjust automatically. So, when a user gets a grade of 80, it means that based on the factors we look at, that user scored higher than 80 percent of the other users that have been graded.”

He concluded by acknowledging that calculating authority on Twitter is not a perfect science, but that Hubspot, who has been doing this for a while, has evolves its algorithm as it learns more about how Twitter is being used.

Twitalyzer

On behalf of Twitalyzer I heard back from its creator, Eric T. Peterson, Chief Executive Officer and Principal Consultant at Web Analytics Demystified, Inc (and
@erictpeterson on Twitter).

Eric says Twitalyzer, which can spit out analytics in dozens of categories, decided that humans know that influence is “something we see and experience every day in our lives” and “that being complex for complexity's sake was a bad idea. So we fixed that.”

With the Twitalyzer 2.0 release his company made changes to the old "influence" calculation and started calling it "impact."

“At the same time we dramatically simplified the influence calculation to look at the two measures we believe best reflected the pure definition of influence (‘causing something without any direct or apparent effort'): retweets and references,” Eric said.

That way anyone tracking their influence in real time will find “the more you invest in Twitter as a communication medium, the more influential you will become,” he said.

And while he personally does not see influence on Twitter as some kind of contest “if you're a business person paid to Twitter and you can't move your Impact score up, something is wrong.”

“This is why we provide recommendations in Twitalyzer Dashboard ... to help our business users understand how they can improve their impact in Twitter,” Eric said.

TweetLevel

On behalf of TweetLevel I heard back from its creator, Jonny Bentwood, Head of Analyst Relations and Strategy at Edelman in London (and
@jonnybentwood on Twitter).

TweetLevel tallies up scores in popularity, engagement and trust to reach an influence score. It does this with a transparent calculation it reveals on its
About page that includes data from such places as Twitalyzer and Twinfluence.

Jonny told me that the key aspect when measuring influence with Tweetlevel is context. Within context the tool examines the “location of the conversation” and the type of influencer (including those that create, amplify, adapt or comment).

TweetLevel doesn't “confuse popularity with importance,” but instead focuses on “micro topics” and understanding who are the key people in those areas. Jonny added: “TweetLevel is a dynamic tool that will alter someone's score depending upon that user's current usage.”

“The tool will continually evolve as Twitter adapts its API and functionality (such as the recent introduction of lists and change in the way retweets work). TweetLevel will take this new data and include it in the algorithm,” he said.

In terms of what TweetLevel's scores mean (on a scale of 1 to 100) Jonny said his tool's score is the equivalent of someone's Google PageRank x 10. A score above 80 is truly exceptional and below 25 is not too good. For the everyday user that means a score over 40 is good, over 50 is great and over 60 is amazing

So what did I learn? It seems that of the three tools I use (for myself and clients) all have their value.

I think anyone relying too heavily on one influence measuring tool is likely getting a skewed picture of influence. Maybe there is a place for an “influence aggregator” that scores each user across multiple influence tools.

What do you think? Does Social Media , and Twitter specifically, need this?

Possibly related post
7 Tools To Find Who's Big on Twitter


Link to original post
Mike Johansson

Mike Johansson

Senior Lecturer at Rochester Institute of Technology and principal at Fixitology, RIT and Fixitology

Mike is a strategist and teacher who helps businesses and students understand and get the most from social media. He currently is a Senior Lecturer in the Department of Communication at the Rochester Institute of Technology where he teaches advertising, public relations and journalism (all with a social media twist). 

See Full Profile >

Comments

RichBaker
Posted on March 1st 2010 at 3:32PM
Great post thanks for sharing! Without doubt my favourite tool is Klout.com. I interview the CEO of the company here; http://www.facebook.com/notes/digital-engagement-and-social-media/klouts...
MatthewRoyse
Posted on March 1st 2010 at 4:07PM
As Rich mentioned above, Klout is another tool to add to your excellent list, Mike.  I talk about the tool in my blog post 10 Ways to Grow Your Twitter Influence.  Also, it is worth mentioning that Hootsuite has integrated Klout into its platform.
chrisaangell
Posted on March 1st 2010 at 4:40PM

Thank you, regarding "TweetLevel", you advise

'the tool examines the “location of the conversation” and the type of influencer'. 

Are you referring to a geo-location? 

I work for NAVTEQ and this is my personal question

 

Mike Johansson
Posted on March 1st 2010 at 5:46PM

Mark Clayson - Thank you for the props. And, yes, I am a believer it what goes around comes around ...:-)

Mark Schaefer, Rich Baker and Mathew Royse - thanks for your comments and the additional pointers

Chris A. - You know that's a great question for TweetLevel;. I read it as the location being whether the coversation was between you and an influential user of Twitter or a low influencer, but now that you mention it ...

chrisaangell
Posted on March 1st 2010 at 9:59PM
Thanks for reply Mike and understood; I'll look more closely at TweetLevel .
Courtney Hunt
Posted on March 8th 2010 at 12:16PM

(I wrote a response yesterday but forgot to log in first, so it may be lost in cyberspace. I'll try to recreate it.)

I am not familiar with these influence graders, but one of the things that struck me as I was reading the piece and the comments was that there may be a bias toward quantity over quality. One of my biggest frustrations with Twitter is that many of the most active users, who have large followings (at least for non-celebs), post a lot of dreck. In addition, they tend to engage in private discussions publicly by not using the DM function or taking their personal exchange onto another platform. In other words, the signal/noise ratio is terrible! By certain standards, however, these folks could show up as great influencers, which of course could reinforce their bad behavior. I would love to see a tool that can discern the signal from the noise and develop quality-focused analytics. Not only would that give a better answer with respect to who's demonstrating influence, but it can also help improve the overall quality of the conversation in the Twittersphere.

Mike Johansson
Posted on April 9th 2010 at 6:21PM
Courtney - if you haven't tried Twitalyzer, I would recommend it. It does measure the noise-to-signal ratio yo seek and does it well, I think. I agree that in these relatively early days these tools could all use some improvement (in some cases a lot), but I think you might find Twitalyzer useful. Thanks for the thoughtful comment.
Courtney Hunt
Posted on April 10th 2010 at 9:42AM
Thanks, Mike. I've added Twitalyzer to my list of "Twitter To Dos." Though I've been on Twitter for about a year, I haven't invested a lot of time on the three accounts I have (1 personal - used for receiving news from various sources; 2  business - used to share resources, information, and updates) beyond following a handful of people/organizations and posting updates. There are other platforms and channels that produce much greater benefits, so I tend to focus my energies there instead. One day Twitter will get more of my attention ...