Blinding Me With Science

Conventional wisdom holds that “if you can’t measure it, you can’t manage it,” so we have metrics to help us manage our businesses.

And then there are Twitter-related metrics.

Meeyoung Cha from the Max Planck Institute for Software Systems looked at data from all 52 million Twitter accounts and determined that:

“The number of followers a Tweeter has is largely meaningless. Popular users who have a high number of followers are not necessarily influential in terms of spawning retweets or mentions,” she said. The more interesting question is how should one measure influence, she continues. Unfortunately there is no one easy answer to that, she says. “One would have to take a combination of many metrics, including follower count, mentions, and re-tweets. However the hard part is figuring out the relative importance of the component metrics.”

Cha is spot on that follower count isn’t important. But she’s wrong when she says that the hard part of measuring influence is “figuring out the relative importance of the component metrics.”

The hard part is figuring out what influence is. When you figure that out, then you can start arguing about how to measure it.

Social media analytics firm Sysomos conveniently avoids defining what influence is, and has developed a metric it calls the authority ranking: A score between 0 to 10 – with 10 signifying someone with very high reach and influence.

Social media “heavyweights” Chris Brogan and Jeremiah Owyang have an average follower authority (an “AFA” if you want to sound cool) of 4.0 while Jason Falls’ AFA is 4.8, and Scott Stratten’s is 4.6.

I guess we’re to conclude that Jason and Scott are more influential than Chris and Jeremiah.

If they want to raise their AFA, Chris and Jeremiah can cull through their list of followers (139k for Chris, 65k for Jeremiah) and block those with a low AFA. And then, going forward, only allow people with a high AFA to follow them.  I can’t think of a bigger waste of time, or stupider thing to do.

I could be off-base here, but to me, influence is about shaping how people think and/or act, wouldn’t you agree?

If you do, then how in the world can you measure influence simply by looking at follower count or follower’s follower count, retweeting activity, or mentions? What does any of that have to with influence?

Answer: NOTHING. Those “metrics” have nothing to do with influence.

I can’t tell you how many times I’ve DMed someone who has tweeted a link and asked “You believe that load of crap?” only to receive the reply “oh, I don’t believe it — I was just passing on the link.” If they don’t believe it, then they really weren’t influenced, were they? Nor are they being influential, because, apparently, they’re not trying to shape anyone’s thoughts or behaviors.

Most of these Twitter metrics are just pseudo-scientific stabs at establishing a system for score-keeping.

Don’t get me wrong: I’m not suggesting that you stop bragging about your follower count, influence ranking, or AFA score. Anything that helps you deal with your personal insecurities is OK in my book. But don’t try to blind me with your science. It’s not working.

Technorati Tags: 


22 thoughts on “Blinding Me With Science

  1. Well put Ron. All the followers and “connections” in the world don’t amount to anything if we are not forming true relationships. Relationships where we hold each other accountable, dream together, and challenge each other.

  2. Janine: You just opened up another can of worms. While I was focusing on the inability of these emerging metrics to measure influence, you bring up the subject of “relationship”. I believe — and maybe I’m deceiving myself — that I use Twitter more to develop and enhance personal relationships than I do to “influence” people. If all these bogus metrics aren’t even good at measuring influence, then you can be sure that they’re no where close to being good at measuring “relationship development”.

  3. Pingback: Tweets that mention Blinding Me With Science « Marketing Tea Party --

  4. Whilst reading I was reminded, albeit vaguely, about a problem in linguistics. There are scientific ways that linguists can track words, pronunciations and turns of phrases through social groups. As I recall, a change in the language is usually attributed to a very few individuals, often women (traditionally in the village/the social communicators). The dropping of R’s in Boston or the nasal twang from the mid-west etc., etc., especially the mutations, are reputed to be started by very strong/influential people who are followed in what we’d call a viral manner.

    After I read all of today’s tweets I might try and find some reference to the research…

  5. With the exception of celebrities, those with the most followers on Twitter are multi-level marketers, get-rich quick/online scammers and so-called “social media experts” — all people who use “volume” as part of their circular and self-reinforcing schemes.

  6. JP: Dude! It’s my circular and self-reinforcing schemes that help keep me sane. I gotta drink me some of that Twitter influence kool-aid.

  7. How do we influence people? Does one say – I’m going to go out and influence people? Maybe. But to an old salesman like me – it’s all about relationships. I ain’t gonna get anywhere if I don’t develop a relationship first. I don’t worry about the sale. It’s the process and relationships that get me there. Seems a bit cold to try to measure that.
    Twitter has been great for me. Gotten to know lots ‘o folks…learning a lot too. (FYI – I’m not overtly selling anything on Twitter.)

  8. Ron –

    Its well known that you have 7 twitter accounts and you tweet and retweet yourself, and #FF yourself. As I recall the accounts include:

    mydaughtersarefantastic, and

  9. J-Eye: I can’t believe you outed me like that. Yes, it’s true, I have those accounts, but truth be told, they were all patterned after your Twitter accounts:

    MyKidsHateMe, and

  10. This is all very true… but everyone knows you need to get a high enough Social Media Influence score to qualify for getting free stuff! What else is social media for? 😉

  11. Well put! I don’t think the number of followers says anything at all about how influential a person is, just how “popular” and even that is a term that I use loosely. I love that you have even asked people if they believe what they retweet. That was a real eye-opener.
    Of course, companies won’t necessarily understand this. They want those metrics and measurements and they want you to tell them what they mean. Otherwise, what are they paying you for?

  12. Nice. But still..isn’t Re-Tweeting a reasonable measure of influence? I suppose some blindly RT but just because they may disagree doesn’t mean they weren’t influenced
    and I RTed your piece and am influenced (so there!).

    But no, it ain’t science. It’s marketing. At best a dark art.

  13. Ron, nice post. Now that we know how not to measure influence, how do we define and measure influence? Knocking ideas down is easy. What’s the solution?

  14. Helen: So funny that you’d mention that. Yesterday I heard that Virgin Air was giving discounts to people with a high Klout score. Blows the whole premise of my post out of the water. Damn them.

  15. Bob: I agree w/ you that RTing is a measure of influence. You probably got here as a result of Avinash’s tweet. Thanks to HIS influence, a lot of people clicked on the link to this post. I think my issue w/ many of the so-called influence metrics is that they: 1) don’t define what influence is, 2) try to turn something very objective (i.e., their algorithm) into something scientific, and 3) include what is easily measured (ie., average follower authority) rather than what should be measured.

  16. I wouldn’t defenestrate the proverbial baby here with the waste water. It could be that the best metrics just haven’t been identified.

    What if we determined several points to apply to each Twitter user:
    1. Tweets per time period. Someone who is more active would have a higher score
    2. Apply a coefficient based on readership vale (this would be complex, because THOSE users value would be based pn several points think Google site rank applied to individuals)
    3. Retweetableness – give extra points to users that author tweets that end up getting retweeted.

    It wouldn’t be a simple task- but I have no doubt that it could be done, and done well.

  17. Jason: Thanks for commenting. Yes, knocking ideas down is easy. But I’d like to think that I’m making a logical, rational, and bottom-line oriented case for knocking it down. In order to tell you what the “solution” is, i.e. how to measure influence, I have to ask you a couple of questions: 1) What does influence mean in your business? 2) Why is it important? and 3) What are you trying to fix or improve by measuring influence?

    This gets at the issue I have with these pseudo-scientific metrics. A good metric tells you how you’re doing, a really good metric help you figure out what actions to take to improve your business. These twitter metrics do neither. They take a bunch of behaviors — i.e., retweeting — and slap the “influential” label on them.

  18. Ron –

    As suspected, there is an entire field of anthropology and linguistics that analyzes “social networks” and measures, among other things, the influence of members.

    Here’s but one Wikipedia link:

    I believe there’s an academically researched answer to what is influence, how can you measure or observe it, etc.


  19. J-Eye: Thanks. I particularly like the eigenvector centrality metric: A measure of the importance of a node in a network. I think we can calculate the Twitter Eigenvector Centrality score as follows:

    (# of followers/# of ppl following) * average # of daily tweets * average # of hours tweeting in a day * SII (Self-importance index)

    My TECS is really really high.

Comments are closed.