Influence: What Are Tools Like Klout Really Measuring?

This is part of a series of posts based on excerpts from The Realtime Report’s Guide to Influence Measurement Tools.

For marketers, PR professionals and customer service teams, personal influence measurement tools can save time and help facilitate business decisions.  Tools such as Klout, PeerIndex, Kred and TweetLevel are being used by brands to rank the relative importance of customers and prospects, prioritize customer service responses, and identify groups of influencers to target with perks and product sampling promotions.

But what are these personal influence measurement tools really measuring?  Are they really an effective way to understand which of your customers are more influential?

It is easy to understand influence as a concept; if you can get other people to do something, you have influence. But it’s not at all easy to define how you would measure influence.  As Nathan Gilliatt has pointed out, there is no such thing as a “unit of influence” – an observable, measurable event that reflects influence.

Defining a Model for Measuring Influence:  What You Measure Is What You Get

For a measurement model to be useful and credible, you need to understand the measurable events on which the model is based.  Unfortunately, most of the tools that claim to measure online influence are not very transparent about what it is they’re actually measuring.

Personal influence measurement tools measure and analyze your activity on social networks, and how other users interact with you, by examining variables such as:

  • Reach: the number of fans and followers you have and, in some cases, the reach of your fans and followers
  • Follower ratio:  how many people follow you and how many you follow
  • Activity: how frequently you post status updates
  • Engagement by others with your content: how many re-tweets, likes and comments your posts get; how many different people engage with your content; how far your content travels
  • Your engagement with other people’s content: how often you comment, retweet or interact with other people’s content
  • Semantics:  keywords and topics that appear most frequently in the content that you generate, and which gets the most reactions from others
  • Your network: the users that follow you are most likely to engage with your content, and the users with whose content you are most likely to engage
  • Ratings from other users:  some platforms have added the ability for users to rate each other’s expertise or authority in certain topics, adding a human layer to the score

These tools can only measure what they can see.  For example, a high Klout score indicates that the user is active on the social networks to which Klout has access, and engages in activity that Klout has defined as being important.

For business users, this means that the scores and other data generated by personal influence measurement tools are useful only in evaluating your customers’ activity on the social networks that are being measured.  If a customer is not very active on Twitter and other social networks, personal influence measurement tools will not necessarily give you an accurate view of that customer’s relative importance to your business.

Show Me the Model:  What Are Personal Influence Measurement Tools Actually Measuring?

Influence measurement tools like Klout or Kred use words like “trust,” “authority,” and “impact” to describe what they’re measuring.  But what actual behaviors and activities are they really measuring?  In the table below, we look at how each describes what it measures about social media user activity five key areas:  activity, reach, network influence, engagement and content.

Each tool emphasizes different types of measurement strategies in the description of their model—but few of the influence scoring tools describe their model with enough precision to let you understand what the score really represents.

One notable exception is Kred, which provides activity reports that let you literally see each specific action recorded by the system, and how many points that particular action contributed to an “Influence” or “Outreach” score.  The site includes detailed descriptions of how the scores are calculated, and is very transparent about how its model works.

TweetLevel is also extremely descriptive about its scoring model.  PeekAnalytics is relatively straightforward in how they describe the data they measure.  PeerIndex loses a little bit of ground in that its “Authority Score” is based on a metric tied to an individual’s Authority in specific topics, and it’s not entirely clear how those topics are assigned. Klout offers very little detail about its model at all.

PeerIndex and Kred both sort users into content-specific buckets, and then measure the level of “authority” a user has in that topic or community.

TweetLevel does not score topics, but it does say that its measurement model includes “thought leadership and idea origination”—this likely means that a score increases if a user frequently posts original content that is retweeted.  “It is important to remember that TweetLevel unlike other tools places far more emphasis on people who create ideas as opposed to merely spread them,” says TweetLevel developer Jonny Bentwood.  “This is why individuals who normally would have a low score on competing solutions can have a high number here,” he told me.

How Each Tool Describes What Its Influence Score Measures:

The table below shows what each of the tools say they include in their scoring model.  The information is based only on what the vendors are disclosing on their web site.  No information doesn’t mean that the activity or metric is not included in the score, just that the vendor does not provide information on how that element fits into their model.

Klout Kred PeerIndex TweetLevel
Activity Assigns 10 points for actions like replying to people and retweeting content How much you do that is related to the topic communities you are part of Frequency of updates, @mentions and @replies; nr. of retweets; broadcast to engagement ratio; involvement (commenting, asking questions, etc.); velocity (the speed of tweets or retweets)
Reach How many people you influence The number of people who are “impacted by your actions and are receptive to what you are saying” Popularity, based on followers and lists
Network Influence The influence of your network You receive more points if someone with a large following does something for you Influence (authority of followers, name pointing, etc)
Engagement  “How much you influence” the people in your network Engagement level; trust (retweets and references)
Content Kred’s Community Scores are based on “Influence and Activity” with other people that Kred has assigned to that community “Authority” in eight benchmark topics, which are used to generate the overall Authority Score Idea origination/thought leadership

The only tool that describes how it measures each of the five areas is TweetLevelPeerIndex and Kred both include topic- or community-specific scores in their overall influence score.  Klout’s description focuses on reach, network influence and engagement—but that doesn’t mean that the other factors don’t play a role, only that their description of methodology doesn’t include them.

Given the lack of transparency about the scoring models, it is very difficult to know exactly what an influencer score represents.  The best we can do is to say that it measures a variety of elements of a user’s activity on certain social networks.  This is useful—but it’s important for marketers, PR professionals, customer service experts and other business users to understand what the scores do and don’t represent when deciding how much weight to give them.

The Realtime Report's Guide to Influence Measurement Tools

Does your business use personal influence measurement tools like Klout or PeerIndex to score and rank customers?  How useful are you finding the results to be?



To learn more about influence measurement tools, check out The Realtime Report’s Guide to Influence Measurement Tools, our detailed analysis of all of the tools mentioned in this post.