Note: This is a continuing excerpt from what I teach my social media marketing classes on introduction to metrics and measurement…
Measuring Influence: Art or Science?
Wouldn’t it be great if an organization could identify people who were highly influential to their prospective users and customers and reach out specifically to them? This could be the key to “Word of mouth on Steroids.” Marketers have struggled for years with the idea of influence over prospects in choosing spokespersons and endorsers of products. Celebrity endorsers are measured with a Q-score, a metric that combines public awareness of the celebrity and the appeal of the celebrity. The Q-score is often considered a measure of likeability. However for many products and services, the endorser’s likeability is not enough; source credibility is also important. Source credibility[i] is seen as vital for products that are major purchases or perceived as complex. Source credibility, as usually measured, comprises expertise and trustworthiness; the spokesperson would be likely to know and understand the product or service and be telling the truth about it.(Hovland and Weiss 1951)
Multiple services are trying to measure the “influence” of social media participants.
At the time of this writing six of the most prominent are Klout (http://www.klout.com), PeerIndex (http://www.peerindex.com), Kred (http://www.kred.com), TweetLevel (http://tweetlevel.edelman.com/), TweetGrader (http://www.tweetgrader.com), and Twitalyzer (http://www.twitalyzer.com). These services attempt to measure a social media participant’s influence through the interactions with others in social media and the participant’s social media community. For example if after a participant tweets or posts a message on Twitter, a follower clicks on a link in the message, re-tweets the message, or “favorites” the tweet, that action would be seen as an indication that the follower values the message and the originator of the message has influence. The number of such actions is weighted by the number of followers or friends that the participant has so not just the number of such interactions but the percentage of followers who interact is measured. In fact, for three of the five services mentioned (the exceptions being TweetGrader and Twitalzer) the participant does not benefit from simply having a large number of followers.
An Overview of the Influence Measurers
There are a growing number of organizations measuring the influence of participants in social media. I periodically check my personal score on five of the six leading influences measuring services mentioned, Klout, PeerIndex, Kred, TweetGrader, and Twitalyzer to review their analysis of my social media activity. A quick overview of each follows, more information and up-to-date information is on their websites.
The two most prominent services, Klout and PeerIndex, use secret algorithms to analyze interactions between social media participants and their communities resulting in a single index measure on a 100-point scale. PeerIndex displays a graph on its splash page showing a participant’s influence in a “topic footprint” comprising eight broad topics, including the Internet, science, leisure, and business. As noted, the algorithms that determine the measures are secret, but the companies have indicated that all other things equal, a participant will have a higher score in these indexes if they engage a community as evidenced by having: a high percentage of Twitter followers react to their tweets by “retweeting,” “favoriting,” or commenting on a tweet or by clicking on a link in the tweets; and/or having a high percentage of Facebook friends like, comment, or share their posts on that platform. Participants’ scores also benefit from a high level of activity and the high influence measures of people in their communities. Twitter CEO Joe Fernandez has recommended interacting with people who have high Klout scores to improve a Klout score. Median Klout and PeerIndex scores are below 20 and the 99th percentile kicks in for scores in the 60s so scores 90 and above are rare. Until a major revision of the algorithm in August 2012, Klout was often criticized because:
• The only 100 on the Klout influence measure had been achieved by teenage heartthrob Justin Bieber.
• President Obama had Klout influence of 94.
• Reality celebrity Kim Kardashian had a 91 Klout measure.
Kred is one of the newer influence measures; its motto is “We all have Kred somewhere.” Kred advertises its transparency: participants are awarded points for various activities. The activities are generally the same engagement activities that Klout and PeerIndex focus on, including retweets, favorites, comments, as well as for follows, and the participant can actually see the points as they are awarded. The Kred site includes an activity breakdown which shows the points that are awarded and a chart of distribution of scores, which made the estimation of percentiles in Table 5.3 easy to estimate for Kred. Kred does not weight the scoring for the influence of the participant interacting, so someone trying to game a Kred score need not focus on social media participants with high Kred scores. Kred, like Klout, allows participants to award each other recognition and like both Klout and PeerIndex, it is developing ways to reward and honor those with high influence scores. In addition to the influence score, Kred also offers a separate “outreach” score to indicate reciprocal interaction by a participant. Kred offers the option to upload “offline” accomplishments in education, career, or interests which a team will access for additional Kred. Klout made a new release in August 2012 that aimed to be competitive with the transparency of Kred and at the same time including offline accomplishments and +K grants from other social participants in the Klout score. Now social participants can study Klout “Moments” and Kred “Story” to gain insight into their social influence.
TweetGrader and Twitalyzer measure interactions but also include the results of direct activities such as number of tweets and number of followers that critics claim can be too easily manipulated. Some participants have been known to announce they will follow everyone back, automate followers, or even buy followers from services to build follower count. Similarly some auto-tweet or post for RSS feeds to create activity. Both TweetGrader and Twitalyzer show influence scores as percentiles, although TweetGrader allows rounding to show some participants with 100 scores. Twitalyzer has a dashboard of multiple measures including the Klout and PeerIndex breakdowns, so it is a useful site for a quick analysis. TweetGrader has some interesting services including local rankings, so that on their site one can view the top fifty TweetGrader scores in a given city or region.
Table 5.3 shows the approximate percentiles that correspond to an influence score from the five influence measurement services. These are guidelines only; in the case of Klout and PeerIndex, the percentiles are my best estimates at the time of writing this chapter.
Table 5.3. Comparative Influence Scores—Approximate Percentiles for Scores
TweetGrader™ and Twitalyzer
Source: Author used information from http://www.klout.com, http://www.peerindex.com, and http://www.kred.com as well as http://www.twitalyzer.com to make estimates; only Kred and the two based on percentiles were fully transparent. (May 2012)
There are multiple services providing measures of online influence beyond the measures discussed already, Klout™, PeerIndex™, Kred™, TwitGrader™, Twitalyzer™, and TweetLevel™. In Appendix B of ROI[ii] (Schaefer 2012, p. 189–93) known influence measuring services are briefly summarized by methodology, use, and descripton, including web addresses. Schaefer includes Klout, PeerIndex, Kred, TweetGrader, and Twitalyzer as well as the following (alphabetized):
• Back Tweets
• Empire Avenue
• My Web Career
• Social Business Index
• Social Mention
• Sprout Social
• Square Grader
• Twitter Counter
This is part #3 of an excerpt about “Metrics” from an early draft of a text for teaching Social Media Marketing. Please do not copy without the approval of Flatworld Knowledge and Gary Schirr. I welcome thoughts on omissions, additions and corrections!!!
[i] Hovland, Carl I. and Walter Weiss (1951), “The Influence of Source Credibility on Communication Effectiveness,” Public Opinion Quarterly, 15 (Winter), 635–650.
[ii] Schaefer, Mark W. (2012) Return on Influence: The Revolutionary Power of Klout, Social Scoring, and Influence Marketing, New York, McGraw-Hill, 215 pages.