Social Media Case Studies, Brand Communities

The Fight for the Future of Influencer Analytics

Jay Baer Blog Postnew klout interfaceBig news last week as Klout announced a significant overhaul of its influencer identification platform. Several new features are being rolled out to improve accuracy and add transparency to the much-discussed and oft-maligned system.

The most notable changes are:


  • An increase in ranking factors from approximately 100 to 400, meaning there are vastly more activity and behavior types comprising your Klout score.
  • The first forays into trying to account for offline influence, using Wikipedia pages and mentions as an important signal (since they are harder to fake and game than most social media).
  • Significant increases in transparency, as a new social interactions ticker shows what you published and who it impacted. (interesting to note that Twitter’s new “you must make your feeds look like” edict last week may have made Klout’s new interface DOA).

Klout (and its main competitor, Kred) are trying to determine how influential a person is in comparison to other persons, and as a secondary data point how influential that person is about particular topics. Both systems mix and combine a lengthy list of circumstantial influence ingredients and then knead to produce a unified mathematical dough that assigns a single, 1-100 number (1-1,000 for Kred) that serves as a literal and figurative measuring stick of influence.

Fair enough. Very easy to use. Very easy to understand. Very easy to integrate into other applications. But as I think about the future of influence, I wonder if maybe the future is in specific influence scores, not general ones?

Your Influence is Channel Specific not Evenly Distributed

I’ve said this before, and I’ll keep saying it. The biggest problem with this entire field of inquiry is that it often confuses “influence” and “audience”.

Influence isn’t about eyeball aggregation, and what businesspeople care about (or should) is how effective the person in question is at DRIVING ACTION amongst the people to whom they are connected. Using influencers to solely drive awareness is as cost-effective as a Paula Deen fitness camp. The key to effective use of influencers is their ability to cause behavior, and that ability varies not just by topic, but even more so by platform. The new Klout is improved on this front, as it overweights action (clicks and other engagement) and underweights activity (tons of pointless tweets).

Nobody has evenly distributed influence. Nobody. Whether it’s Twitter, Facebook, Instagram, Pinterest, a blog, YouTube or any other corral, everyone has places where they are more connected and more capable of moving the behavioral needle. Sometimes this is evolutionary, stemming from how and why and when you joined/embraced a particular network. Other times it’s by choice, as with Chris Brogan publicly eschewing Linkedin and Facebook in favor of Google + and Twitter alone (which is one of the reasons I now have a higher Klout score than Chris, which of course does not reflect reality).

Klout itself is visually acknowledging the divergent and network-dependent nature of influence with their new pie chart that shows where influence is accrued. In my case, 49% of my total Klout is via Twitter, with Facebook at 40%, and Linkedin at 5%. They of course have always possessed this data, but are now showing it as part of the tool.

Jay Baer Klout

Klout says that I am more capable of getting people to click, et al on Twitter than I am on Linkedin. That is true, and I can look at the analytics from my Buffer and Argyle Social accounts to prove it. So if a company is evaluating me as a potential outreach target, shouldn’t they want to work with me only on Twitter and Facebook? Is focusing on channel influence more efficient and effective than worrying about baseline influence?

The Rise of Channel Influence Analytics

If you believe channel influence is important, there is an emerging and expanding crop of analytics companies that provide excellent insight into the situation. 

Twitter Influence Analytics

Plexus Engine - Top Twitter Users on the Topic of ruby

Plexus Engine report excerpt for “Ruby”

On the Twitter side, you’ll find Plexus Engine (currently in beta) which instantly gives you a list of topical influencers and allows you to create a list, custom search engine, and blog feed in a single click. It even has “flash cards” in the mobile version to help you learn who’s who on Twitter for a category.

You’ll also find Peek Analytics, which provides exceptionally deep data on individual Twitter accounts and hash tags. Very useful for head-to-head comparisons. For example, using Peek I can determine that my friend Jason Falls has 3.1 times the expected concentration of public sector Twitter followers, compared to my 1.6 times the concentration. This could be because Jason worked in higher ed for many years, and my own government career was just 4 months in duration and happened pre-Internet. Or maybe he’s just a .gov darling. Don’t know the cause, but Peek Analytics shows you the deep data around income, education, age, and location.

I’m also interested in Brandfluencers which ties Twitter data to your Google Analytics account to determine who in the Twitterverse is driving the most visits and pageviews to your website. From the perspective of focusing on ACTION not just audience, it’s a simple yet terrific tool. In the past month, the top drivers of traffic here to Convince & Convert (other than me, naturally) are Lisa Barone, Mark Schaefer, Unbounce, and Radian6. Thanks all.

Pinterest Influence Analytics

Pinfluencer Top Pins by Revenue

Pinfluencer Top Pins by Revenue Report

Given many pins’ inherent tie to products, e-commerce, and potential revenue, the category of Pinterest influencer analytics is fascinating and booming.

I’m a big fan of Pinfluencer, which provides shockingly valuable data to its e-commerce and retail customers (like Zappos) about which pinners are driving the most re-pins, clicks, and even hard revenue. In fact, Pinfluencer CEO Sharad Verma tells me that some e-commerce companies are using real-time Pinfluencer data to help determine which products to feature on their home page, and even product inventory levels.

Chief competitor Curalate offers similar functionality.

Blog Influence Analytics

One of my favorite online tools of any type, and software I use frequently with our agency clients is GroupHigh, which I believe to be the best blogger search and contact tool available. If you need to find influential food bloggers in Tampa, Florida and then create a database with a record of when you contacted each of them, this is the tool.

Lots of other software (Cision and Vocus, in particular) offer some form of blogger search and outreach capabilities, but GroupHigh is purpose built for bloggers (no traditional press included) and is a snap to use.

Depth vs. Breadth of Influence Scoring

That list of tools is just scratching the surface. There are many, many more with others appearing weekly – or so it seems.

I’m not a Klout hater. Quite the opposite, actually. (Read my post “why critics of Klout are missing the big picture“) But if we really care about behavior, not the ability to drive awareness, perhaps Klout would be better served by creating several network-specific measures of influence instead of a single score influence gumbo?

Would you interested in a separate Klout or Kred score for Twitter, Facebook, Linkedin, Instagram, and so forth? Or is the one-score-fits-all alternative that currently dominates the influence discussion the best option? I’m continuing to ponder this question, and your thinking is most welcomed below.

 (disclosure: I am an investor in Plexus Engine and Buffer, and Argyle Social is a sponsor of this site)

Facebook Comments


  1. mftracy says

    @jaybaer Agree! The score ideally should be @ ‘network’ level and eventually aggregated by data on ‘demos’ , topics/subjects, etc…

  2. eswayne says

    I can’t but help but look at all this fervor over influencer analytics and see it as just version 1.0 of something much deeper.  We all agree that influence = action, but I can’t tell you what actions are successful for you until I have access to YOUR data.  Each and every business has its own success metrics (although many fall into common categories like “eCommerce purchase”), and therefore each business has a different definition of successful “influence.”  I think the end game is something almost more like a meteorological model – we feed all of these wide range of factors from Social, Web, Mobile and more into a single model (or set of models) that help us predict which actions we can take that will move the needles we want to move in the future.  The future is more Social Freakonomics than Social Analytics. 

    • says

      eswayne Excellent, excellent comment. Many thanks. Indeed, if influence = action, then your definition of successful influence will differ from mine. That’s a geometric leap in the type of analysis necessary, and is really in the field of predictive modeling. But we’ll get there eventually. I’m certain of that. 

  3. says

    Good stuff, Jay. As the murky waters of diminishing returns on so much marketing efforts begins to clarify, we are all seeing big changes. With information that ultimately lets us commit our marketing resources to hard evidence, and not hype, we can all get back to basics. Still, there is a part of me that wants this holy grail to fall short of instant marketing gratification. I know, call me a marketing heretic. But what I really mean by it is the notion that when we let our tools, analytics, and expectations of ROI drive less than ideal behavior on our part, I think it does two things. First off, it skews the results of our analytics divination (and this is something no marketer ever thinks about measuring outside of campaign success). And second, marketers continue to avoid direct interaction with their customers. I wrote about this last year, back when Two Bananas was still kicking it. Here is a link to the article.

    • says

      freighter True Marty. The more we focus on the data, the more formulaic and distant our interactions become. No question about it. Not sure that genie is going back in the bottle, however. 

  4. geoffliving says

    Amen: The metric confuses “influence” and “audience”. These are great metrics for reach, but not actual actions and how decisions are made.  These get back to relevance or as Danny Brown mentioned the situation, and who we actually talk to as we consider our ideas.  Influencers or media outlets have a role in that, but I think further away from the decision than we’re led to believe.What drives me crazy about this conversation online is that it completely ignores all of the research that has been conducted in this field. It’s oblivious to the science of networks.

  5. jevgenijs says

    Hello, Jay!Many thanks for mentioning Brandfluencers! Glad you like our product. As you might know, is the product of Campalyst, which we created to showcase our conversion measurement technology.Using Campalyst one can also identify “brandfluencers” based on the referred online revenue and other types of conversions (sign ups, registrations, etc.). Would love to hear your opinion about bringing this functionality to!Thanks!Sincerely,JayCEO and co-founder

  6. JasonMillerCA says

    Great post Jay! Nice breakdown of the tools available. As a marketer at a startup, my challenge is finding the time to dedicate to any one of the tools mentioned above. Instead, I simply build a list of who I consider to be influential based on topics and industry keywords, then I simply pay attention to daily interactions, reach, and post frequency. This is pretty easy for me as it’s what I do daily; monitoring trends and checking into social streams several times a day. I think it’s remarkable how accessible many of these influencers are and I think that many times it’s more of the approach rather than the tool that determines success with an influencer outreach strategy. So although I don’t check Klout on a daily or even weekly basis, it is nice to check in every once in a while to see the data that backs up my initiatives. It would be great if Klout or Kred could show the times that these influencers are most active in their communities, as I believe that is when they are most likely to be respond to messages or comments.Again, great post. Sharing it now….Jason Miller – Marketo 

  7. Lisa Larranaga says

    Hi Jay!
    Influencer Analytics is such a huge discussion and we’re glad to be part of the conversation. Thank you for including us in your round-up but more importantly, thank you for dissecting influencer analytics so thoroughly. Look forward to coming back to this post as I don’t think I can digest it all in one read! Hope your day is going well :)

  8. says

    What a thought-provoking post, Jay. Influence measurement and analytics is still an emerging market, so it’s interesting to see how trends and tools will continue to emerge in this space.I don’t think the question is as much about whether we need individual, all-in-one scoring vs. platform-specific scoring. Rather, it’s about finding influencers who drive action within a particular context. Individual influence only gets you so far. It’s when you can find people who showcase influence by driving action on a particular topic, that you’ve got something worthwhile. When it comes to analytics, at Appinions, we give brands the ability to benchmark their efforts and get a real look at a company’s share of influence on a particular topic or versus their competitors. Being able to measure the shifting volume and sentiment of opinions about your brand can be really powerful.

  9. says

    Great post and debate. I think the focus on driving action is definitely the right way to go, but a challenging one as highlighted in other comments. The other word that I would have expected to have heard more in the debate is ‘relevance’. I guess if you can get to the actual actions then you know you are on topic, but influence is only helpful when definitively relevant to your target audience and I think there is more work to do here – is this part of the secret ‘situational’ analysis Danny Brown ? Look forward to that. :) Thanks again for the post, Jay.

  10. says

    Great post again, Jay. We started with just a top-level influence score, that still means something to individual but not much to marketers. We then moved towards context-specific scores that had less importance to individuals but much more value to marketers. This next step of segmenting influence by channel seems a very logical step. So, where next? I’m betting that geographic influence will also be added to the melting pot soon.Chris Arnold (

  11. sardire says

    @marie_wallace @jaybaer Do any of these measure behavioral changes or actions that change behavior ? From what I see NO & this is the key !

      • SeanGolliher says

        @marie_wallace Problem is using values in isolation. Publishing an absolute number, with no units, relative to nothing, is rarely of use.

        • marie_wallace says

          @SeanGolliher Agreed. Plus #influence can’t be isolated from desired action. Click, buy, sell, visit, eat, talk, laugh, cry, watch …

      • sardire says

        @marie_wallace @jaybaer The hype of influence scoring from Klout, Kred is unbearable. Person Recommendation System like Expertise Network.

Leave a Reply

Your email address will not be published. Required fields are marked *