The measurement of online influence is in its very primitive stages. Like any measurement, people change their behavior in order to meet the standard. In this arena, as it is everywhere, “You get what you measure.”
Klout measures Twitter activities and is weighted towards volume in followers and tweets. mPact measures the impact of the people who sign up for the service. Our approach (using Traackr) tries to capture the entirety of web and social media influence. We look at the entire social graph and the potions of the web covered by list specific keywords to discover the intricacies of influence.
This is the fourth iteration of the overall HR Influencers list. It’s the 13th automated HR list created to date. Each step of the way introduces new understanding and nuance. (Earlier lists, like this one, explain how we calculate the influence measures)
For this round, we experimented with the keyword cloud. It’s become apparent that the lists are vulnerable to SEO style tactics. That means that we needed to change the keywords we used to get more specific. From now on, the keywords will change with each iteration of each list.
It also became clear that the algorithm we use to define influence has a couple of weaknesses.
Sophisticated Twitter users occasionally build their audiences by following huge numbers of people. The unspoken deal is, “I’ll follow you if you follow me.” Since no one can possibly keep up with more than 500 followers, one might cast a suspicious eye on people with huge numbers of friends. In our research, we’re proposing a Friend to Follower Ratio (FFR). The likelihood that someone is influential is probably related to this variable.
For example, if you have 10,000 friends and 10,000 followers, you’d have an FFR of 1. If you had 500 friends and 10,000 followers, you’d have an FFR of 20. The higher the FFR, the more likely it is that you are influential in your circle.
But, since all of the approaches to influence measurement currently rank the number of followers as critical, people work hard to have lots and lots of followers. FFR is probably a measure of tweet quality as well. The more people who follow without a quid pro quo, the more likely that the content is relevant. Hopefully, we’ll get to that by the end of the year.
This time, we added a couple of key players to the bottom of the list. Bill Kutik and Wes Wu are both on the list as a way of explaining the limitations of our process.
Kutik, who many of you know from his role with the HRTech trade show, runs a huge group on LinkedIn. With 5,000 ish members, it is easily the largest forum in HR on LinkedIn. He is an extremely active participant. Unfortunately, LinkedIn is a closed system and data about the people who use it is extremely scarce. Kutik would rank much higher on the list if we had better access to LinkedIn’s social graph.
No one does.
Wes Wu is also on the list. A perennial favorite of the tech crowd, Wes’ aggregate score would put him in the list. We have set a minimum score for relevance that he didn’t meet.
Hopefully, this version of the HR Influencer’s list will give you a reason to think a little more about who is and isn’t influential online.
Meanwhile: here’s the list.
Our lists are compiled with the least amount of human intervention possible. Determinations of influence are made through automated measures. In this edition, my name was pulled out of the list.
The method involves a dramatic (very large) spidering of the web for content related to Recruiting. All of the material that includes prioritized key words (see the image on this page for the keywords) are collected in a single database. It is then compared and contrasted with the data in the social graph.
In order to really quantify the dimensions of online influence, we measure three key variables:
- Reach: A measure of the audience size (number of eyeballs) for each individual. Traffic.
- Relevance: The degree to which content associated with the individual matches a cloud of keywords prepared for the analysis
- Resonance: The number of mentions, inbound links and participation found for each individual.
Keywords
