Skip to content

…of influence and klout

Dear Reader,

Both of you that read this blog regularly know that I’ve been playing around with the Klout API. For a while now Klout has been my new shiny. Last week I started actually hitting their API and gathering info. At first, I just wanted to see what they had available. However, as I started to see the data come in, I started to find out some interesting things. (A little side note, their API is run by the cool folks over at Mashery. /me waves oh hai to Kirsten and Rob)

For amusement purposes only!

First, I think any tool that tries to use twitter to figure out how important someone is in the real world should come with a “For amusement purposes only!” warning label. I’ve played with almost every one that has come out and so far I’ve never found one that is anything more than amusing. Klout falls squarely into that bucket and so I warn you, don’t put any stock in any of the stats I post here, it’s just for fun.

I’m POPULAR…well, maybe…if you stand back and squint

The first thing I did when playing with Klout, before I started playing with the API was see what I could do to increase my score. As of this writing, I’ve moved it from a 54 to a 59 in roughly two weeks just by changing my behavior on twitter. I don’t know what the curve looks like I know that Justin Beiber has a Klout score of 100, neither of these numbers tell us much. I am pretty sure that a score of 60-62 is about as high as I could go. In reality, I’m just not that popular. (That’s ok, I’ve got a fun group of friends on twitter and I’d rather have a small group of friends than a large group of people I don’t know.) However, poking and prodding at the Klout score with my twitter account did make me more curious, so I decided to start playing with the API.

The devil is in the data

Being an old-school database guy, I immediately wrote me a simple script that took the output from the Klout API and stuffed it into a database so I could query it with my favorite MySQL interface, SQLYog.

Once I got a working script and started seeing data in the database, I found something interesting. Klout knew me better than I thought it did. Klout has basically 4 score it computes.

  1. Score – This is the score you see if you go to klout.com.
  2. Class – Any score over 35 gets a class. I am a Specialist whereas Derick Rethans is a Thought Leader. You can see them all on any Klout.com profile, just mouse over the squares in the grid.
  3. Network Score – This is the influence level of your network. If your network isn’t populated by important people, you can’t be THAT important. :)
  4. Amplification Score – How much does your network spread shat you say to the wide area twitter network?
  5. True Reach – This one is my favorite. According to the klout.com web site true reach is:

    True Reach is the size of your engaged audience. We eliminate inactive and spam accounts, and only include accounts that you influence. To do this we calculate influence for each individual relationship taking into account factors such as whether an individual has shared or acted upon your content and the likelihood that they saw it.

All of that you can see just from the web site. However, the API also sends you descriptions of each score, reading them is where I found things got interesting, especially since couldn’t find this text anywhere on the site. What caught my eye was my kscore_description. On the site it says

Cal Evans creates content that is spread throughout their network and drives discussions.

To me, this sounded pretty dang impressive. However, the API added

calevans has a low level of influence.

This is, of course, downright insulting…it’s true, but it’s insulting when an API tell you this. :) So it does know me better than I expected. I started to wonder what else showed up in kscore_description. Reading them I found one to be very interesting.

*** is in the top quartile of influencers and is able to drive conversations.

This is interesting because you would think that this would be reserved for high kscores. If Klout tracks influence, the top quartile should be the top 25% right? (kscore 75-100?) To the contrary, my query on my growing database shows that this designation is given to klout scores from 50.18 – 75.16 It’s almost like this designation is give to the second quartile. (All Klout scores are rounded, so of the scores I have designated “top quartile” the scores range from 50-75. My sampling is approaching 10,000 so I think it’s safe to say it’s large enough so that if there were any 75-100’s they would show up. So while there is no doubt that I am no where near the top quartile, my score is within the range of other top quartile influencers.

A missing keyword

One of the things I personally was looking at this data for was to try and identify the top influencers of the PHP community. Personally, I wanted to make sure I knew them all and if not, reach out to them and meet them. After ~10,000 klout scores identified, PHP does not show up in the keywords section at all. I find that a bit odd. I thought maybe it was just because they were lumping all languages into a single category. This is not the case as my database shows Java, Ruby, jQuery (ok, not a language), heck even Joomla is a keyword, but not PHP. Now I’m not one of those PHP conspiracy nuts that scream because every time an article is written about languages, PHP is left off like some red-headed step child. However, I do find it very odd that PHP is not a keyword. Looking at the klout scores I have cached, I see prominent members of the PHP community that regularly tweet about PHP. I can’t understand how PHP isn’t a keyword for them.

The “K” List

Ok, so my goal was to find the most important tweeters in technology. I could go with straight ksscore but decided that as a measurement, true_reach was much more interesting. So, of the ~10,000 klout scores I’ve looked at, here are the top 25 listed as being in the top quartile, sorted by true reach.

Twitter Handle Klout Score
delbius 75.16
robinsloan 62.06
stop 67.52
JayOatway 71.5
sitepointdotcom 62.7
arrington 68.65
waynesutton 70.16
alexismadrigal 64.49
marshallk 71.08
jolieodell 62.22
jackschofield 68.2
Gartenberg 66.72
RobinGood 69.78
dens 69.35
respres 50.18
mediatemple 69.72
novaspivack 62.29
windowslive 61.62
alex 64.74
KevinMinott 66.26
mjasay 60.7
jobsworth 58.48
Carnage4Life 64.4
Avinio 69.24
nanpalmero 61.27

(now if I could just get a few of them to mention Day Camp 4 Developers… :) )

Next up

Now that I’ve got some data to play with, I’m going back and hitting twitter to get things like total followers, and other data not summarized by klout. The klout score by itself is an interesting metric but I’m trying to figure out more than just a person, I want to identify influential networks. Of course the end-game is to bribe my way into these influential networks with Scotch. :)

Until next time,
I <3 |<
=C=

3 thoughts on “…of influence and klout

  1. Cal,
    On the PHP conspiracy thing.. I will buy you 2 beers, if it’s not related to the fact that there are literally millions of pages a day with .php at the end that are shared on twitter. It seems like something easy to filter through, but it’s not. Or wasn’t for me when I was building geotweet..

    PS: I’m good friends with nanpalmero he’s in your top 25 tech tweeterthingers.. I’ll get him to mention DC4D :).

  2. Pingback: Tweets that mention …of influence and klout | Postcards From My Life -- Topsy.com
  3. @Vid!

    You could be right. However, I would have thought by now that if that *is* the issue, that they could find a way to resolve it.

    WRT @nanpalmero, you rock dude! :)

    Thanks for stopping by and leaving a comment.

    =C=

Comments are closed.