Measuring Online Influence (Liveblog)

by Morgan Siem on August 18, 2011

in Analytics,Social Media

Post image for Measuring Online Influence (Liveblog)

Today’s speaker at the Triangle AMA luncheon is Tom Webster, VP of Strategy at Edison Research. His topic is measuring online influence. He starts with the question, who knows their klout score?

He goes on to make a point that influence is this: Did I compel a change in state?

In order to measure influence you must know:

  • the variables before the measure
  • the measurement gauge
  • the variables after the measure

Measuring influence is much more complicated than some of the tools make it out to be. For instance, influence is only relevant within context. The example that Tom gave was that if he’s going to influence us about airline baggage, he has to establish a context for his influence on that topic, namely that he travels frequently for work.

So what’s the value of influence? An anecdote that Tom tells is that he had one particular message that he sent out to the most influential people he knew (according to Klout). The numbers came back like so:

  • 500,000 impressions
  • 400 clicks
  • 14 entries

What did this tell him besides that any banner ad would outperform his efforts here? It told him that he didn’t understand what influence actually gets you.

He also learned that you sometimes have to look at a variety of tools to get a real sense of who is influential on which topics and with which other community members. It’s just like we look at three different companies to measure our credit scores. None of the tools do everything well, but each have some great data. Tools he mentions include:

  • Klout* – the score isn’t important, but the topics is right on point
  • PeerIndex – pretty accurate on who influences whom
  • TweetGrader - you can drill down to regional influence (one of the first influence measuring tool out there)

*This is the one he gives the biggest thumbs up to.

So, what goes into determining a Klout score or any other online influence score? We tend to think that it’s determined by an algorithm. However, these scores are so far all made based on assumptions:

  • we assume a particular value of an RT
  • we assume a particular value of a reply

However, these are a proxy measure. We don’t know why someone retweeted you – did you infuriate them or please them? Were they making fun of you or supporting you?

So if we go back to the anecdote from earlier, we should point out that 500,000 impressions measures how many people might have seen it. Not how many people actually saw it, how many people will recall it or how many people were influenced on it to take action either online or off.

The thing about Klout scores is that with some businesses they correlate and for others they don’t. For instance, Tom has been an AT&T customer for years and years. But…he HATES them. So, in his case, his satisfaction with the company is not an important metric fro AT&T because he keeps renewing his contract with them anyway. His satisfaction with JetBlue, on the other hand, is highly correlated to his tendency to spend more and more money with them, and recommend them to others.

His advice:

  • Ignore the numbers
  • Use topics as a segmentation guide but not a “ranker”
  • Tie “influencer” as a binary variable to a metric that matters, not a proxy measure (on or off; yes or no; influence or no influence; not 55 vs. 63)

Ended the presentation with the point that he believes that a lot of these tools are on the right track and onto something, BUT “every time you, as a marketer, make a decision based on someone’s Klout score, a unicorn dies.”

A good point that he made in the Q&A was that Klout scores are actually pretty accurate as long as you’re aware that what they really measure is how good you are at Twitter. It can pretty accurately distinguish who is best at disseminating a message on that channel. It does not measure how influential you are at life, just at Twitter. Therefore, it will accurately give you a low score if you are just starting out on Twitter because it takes time to build up your influence.

Who has questions about the topic of online influence? What tools do you use? What metrics do you hope to measure?

Photo by aussiegall

{ 4 comments… read them below or add one }

Arthur August 18, 2011 at 6:11 pm

Yes this is definitely true. Influence is tied to a specific context. An influencer known for marketing isn’t going to bring much action from his followers when he broadcasts a coupon for cooking wares. I’m surprised though, that Tom thinks that the context of an influencer can be understood through twitter at all. 140 characters doesn’t seem like enough room to create interesting quality content for developing user base. Typically, these people are more known for their influence in another property such as blogs or youtube, and their influence carries over to twitter. At the same time, twitters tend to also include conversations beyond the domain of the user’s core focus. For that reason, I think it’s misleading if you try to include twitter into your influence analysis.

We can all agree that relevance and context are critical to defining influence. The difficult part becomes how to find the relevant people, and if you have found them, what communities are they influential in?
Would you agree?


Morgan Siem August 18, 2011 at 8:58 pm

Good points, Arthur. I’ll invite Tom to respond for himself, but in the meantime, here’s my take. I do believe that the context of an influencer can be determined through Twitter. Often, our 140-character tweets are all reflecting of a core set of topics on which we’re influential. For me, for instance, Klout has accurately selected a handful of topics on which it believes I’m influential. These include social media, raleigh and cancer. I can see a clear correlation between those topics and my Twitter conversations. Even just doing a quick once-over of someone’s Twitter account, I can usually gain a great deal of context about the person and get a sense of how influential they are.

Blogs and YouTube are also platforms on which people build grand followings and amass influence, however some of the influencer rating tools are still trying to catch up with those platforms and determine how to integrate social influence there into their scoring systems. It’s a work in progress. Right now, they primarily use Twitter for their ratings because it has fewer privacy barriers than other sites like Facebook.

I think the point made at the end was just that Klout scores are not necessarily scoring a person’s influence, but rather their ability to generate engagement on Twitter. If that’s the key metric, then Klout is an accurate tool for measurement.


Arthur August 23, 2011 at 4:41 pm

I see your point Morgan, and thanks for the reply! No doubt, Klout shows understanding of a person’s relevance an influencer in regard to specific topics in their twitter community. The problem I still see with it (and with using twitter statistics to analyze influence in general) is that there is the possibility that there is little correlation between what a person is really an expert in and what a person tweets about. Of course I couldn’t make that claim without some supporting data, so I did a little data mining yesterday to see how key expressions compared between the top 10 lists for twitterers and bloggers in the social media marketing domain.

You can find my results here:

I’m sure you’ll recognize many of the names of the people I used to develop this data, but in the true spirit of transparency I included the lists of each source I used to create each expression cloud. I also used data for the past 6 months in order to ensure that any currently trending topics wouldn’t create any noise but at the same time the posts would be current and relevant.

The results are what they are. While most of the big words like “social media” and “marketing” were apparent in both, The posts seemed to have a slightly different emphasis. They tweeted quite a lot about social networks, but blogged more about branding and marketing concepts. While we could read into this quite far, but I think that main point from this is that people use different social networks for different reasons and the effects of such are somewhat apparent in this quick exercise.

I think the real questions now are “Which social medium truly portrays an expert’s influence and relevance?” And “Which social medium best helps the influencer drive action from his followers?”. Surely the answer is that it depends, but then it’s up to the person doing the outreach to really determine which social network best represents the influencer. I would think that it’s a poor assumption, and perhaps even dangerous, to assume that an influencer is equally present and influential in every social network they belong to.

Sorry for the essay, your post just sparked my interest. Perhaps I’ll write a blog post about this tomorrow. :P


Brian McDonald August 30, 2011 at 9:20 am

Nice graphic Arthur comparing the blogs to Twitter. I think what drives influence is communication. We don’t know who is an expert or novice on any topic without them stating an opinion. Therefore any scoring mechanism has that fallacy of not including the smartest people if they don’t voice their opinion, recommendation or thoughts.

Regardless of the network choice (Twitter, YouTube, Facebook, LinkedIn, etc.) what Klout and others are trying to do is measure some level of influence based on context, audience size, timing and other factors that relate to the influencer’s message being broadcast to a wider audience. I think one of the best parts of Tom’s presentation was his example of tweeting about the New Zealand earthquake and showing very diminishing returns from broadcast to action.

So we have to take Klout and other measures for what they are worth, measuring influence via social media channels to a perceived audience size based on followers and other measures in order to generate some level of measurement.

To answer your question, “Which social medium truly portrays an expert’s influence and relevance?” The answer depends on the audience and I don’t think you can put the finger on just one answer. Depending on the content, context, and audience it changes or is shared across multiple social networks and platforms.


Leave a Comment

Want to add a picture to your comments here on Media Two Point {OH}? Upload your picture at Gravatar to make it happen.

Your comment may be nofollow free.

Previous post:

Next post: