Social Media Influence, Science or Sham – Part 3
This is the third article in a five part series, and if you missed Part 1 or Part 2 I suggest reading them first so you understand my mission. It is not to assail, grade or score on a definitive or relative basis any of the three primary influence score providers mentioned rather to look more deeply at what they do and how it may affect your life, knowingly or unknowingly.
In this article I’ll be looking at PeerIndex in particular, examining how they produce their scores and reviewing some of the issues I discussed with their CEO, Azeem Azhar, in a telephone interview.
The PeerIndex score is a normalized value from 1 to 100 based on three component scores – authority, audience and activity – in eight topic communities. This approach is interesting from several perspectives and allows creation of a Topic Fingerprint for each user that displays their relevance in specific areas.
The eight topic communities include Arts, Media & Entertainment; Finance, Business & Economics; News, Politics & Society; Sports; Leisure & Lifestyle; Health & Medical; Science & Environment; and Technology & Internet. Based on an evaluation of my own Topic Fingerprint and those of several friends I am impressed with the apparent accuracy of this metric.
The three component scores I mentioned are based on a data algorithm designed to measure a specific value and then these individual scores are amalgamated into your PeerIndex score, represented as a single value. Let’s look at the three components individually in greater detail.
Authority is a measure of trust designed to calculate how much others rely on your recommendations and opinion in general and on particular topics. In essence the more people you affect with your content (measured by mentions, retweets, comments, shares, etc.), within one of the eight topic communities, the greater your Authority score for that topic.
Audience is an indication of your reach, taking into account the relative size of your audience to the size of the audiences of others. This gets a little tricky but in simple terms Audience looks at the number of people who interact with your content relative to the number of people who interact with other people’s content within the same topic. The net effect is your Audience score will be higher if you have more people engaging with your content than somebody who has fewer people engaging with theirs.
Activity is the measure of how much you do that is related to the topic communities you are part of and is also calculated on a relative basis. The calculation takes into account how much activity you create via tweets, comments, shares, etc. and factors in the engagement of your followers and friends via mentions, replies, retweets, comments, shares, etc. Then your activity is viewed relative to the overall amount of activity within the eight topic communities and scored accordingly. Think “big fish, small pond” and vice-versa, so if your activity is a greater percentage of the total activity within a topic your score for that topic will be higher. Conversely if your activity is a smaller percentage of the total your score will be lower.
While it all sounds quite complex, and in reality it is not simple math, the goal is to reduce or eliminate a user’s ability to artificially manipulate their score by constantly evaluating not just quantity but quality, and then looking at an individual in the context of the whole community.
My Topic Fingerprint is a unique feature of PeerIndex, one which I personally like, that displays a quick graphical overview of a user’s topic focus. As you can easily see I spend most of my social media time in the areas of Politics (POL) and Business (BIZ), which is very accurate.
In discussing topic associations with Mr. Azhar, and as the home page states, PeerIndex engineers have been working on a considerable refinement of their methods used to associate a user’s activity with all the topics they track. In that they track almost 8,000 total topics of which 2,000 are currently used in the public interface and they use the last four months of available data for every user, this is no small task but one which is very close to being completed.
In addressing the interface with Mr. Azhar, I pointed out that an area of concern I had was the apparent outdated statistics in my profile such as Twitter followers, Facebook friends and LinkedIn connections. He told me that there are significant new interface features in development and they should be completed within 6 weeks or so, and a more frequently updated interface (Mr. Azhar did point out that the data used to calculate scores is updated daily) is part of those upgrades.
He agreed with me this is an important issue since a user could be given the mistaken impression that the data being used to calculate their score is outdated, which is not the case according to Mr. Azhar. Rather it’s what the current interface is displaying not what is in their databases. I look forward to seeing these upgrades soon and the elimination of this small deficiency.
In addition to associating users with the eight topic communities and individual topics, PeerIndex shows you who you influence and who influences you. The lists are compiled by analyzing the interactions between you and all of your online “friends” and the same person can be on both lists.
As cited in Part 1 of this series there were three issues that I was interested in learning more about as I delved deeper into each application. They are the privacy of my information, the company’s revenue model and their thoughts about the use of SMI scores in areas outside the marketing arena, in particular with regard to their use in determining the qualifications of a job applicant for a given position.
The company specifically says that it does not share publicly identifiable information with 3rd parties unless the user has given specific permission for purposes like participating in a PeerPerk, which is a purely voluntary option for each individual user.
PeerIndex offers a very easy method to control your own privacy settings that dictate what another online user may see from your PeerIndex profile, and you can completely opt out so even your public content will not be scored.
With regard to the other issues I emailed Mr. Azhar two questions prior to our interview and will, for the time being, include them and his responses below without comment. I will hold my opinions and thoughts until Part 5 of this series where I’ll draw all my conclusions and express my thoughts about these two topics.
Please describe what is your company’s revenue model. In short, what do you do with your data to bring dollars in the front door or, lacking any current revenues what will you be doing to create them?
We make money through influencer marketing. We have two products. The first product is PeerPerks – influencer marketing programs – to help companies reach topical influencers relevant to them; topical influencers in the ‘magic middle’ who will end up creating objective discussion and awareness around companies they care about
The second is to use our data programmatically to help companies make better decisions around market segmentation and customer engagement. We believe social data appropriately handled is often better for segmentation that gross demographics – and in some cases better than past spend behavior.
Please cite your company policy and personal attitude about the use of your company’s influence score outside the marketing arena and being used in circumstances such as benchmarking job applicants or determining the value of one applicant over another based on their social media influence score.
Any metric is useful in the right context. A managers Sharpe ratio is relevant in the context of other data about a fund manager, not sufficient on its own. Nor would it be useful if you were looking to build a basketball team of fund managers – then you wouldn’t want to judge them on their Sharpe ratio but rather on their ability to play basketball.
It’s naive to use a single dimensioned influence metric on a CV.
You might want to use topical influence scores for some part of an assessment but in context of other data.
AUTHOR’S NOTE: I would like to thank Mr. Azhar for his time and willingness to be interviewed and for his responses to my written questions. For more information about PeerIndex I suggest perusing their blog.
This is the third article in this five-part series and you can follow me on Twitter (@Tom1247), send me a friend request on Facebook (Tom Dougherty), subscribe to The Right Sphere or my personal blog to get first notice of the next article being posted.