Klout, a service that says it "measures influence based on your ability to drive action" online, is trying to be the new standard by which people and organizations can measure their online impact. And in his blog today, Gillin points out the many flaws in Klout's approach and methodology, such as the limited transparency into how Klout generates its scores and the fact that having a Klout account will raise your Klout Score. (Check out his blog, The Trouble with Klout for more.)
I've discovered two issues with Klout:
1. Klout's algorithm identifies the top three topics you are most influential about -- but Klout is not always accurate. According to Klout, I'm influential about three topics I don't remember tweeting about:
What that says is that it's easy to manipulate the Klout algorithm by tweeting a lot. Doesn't matter if no one else replies or retweets what you've said. I've been able to boost my Klout by posting more frequently -- talking about four or five times a day, not anything like 9,156 tweets. (The converse is true, when I tweet less frequently; my Klout score always drops over the weekend, when I tend not to tweet.)
I understand why my score will drop during periods on inactivity, but just posting a lot should not be a way to boost a Klout score. If someone posts a lot in a short time, and no one responds, they shouldn't see their Klout score increase. More posts do not mean you're more influential. Also, on an interesting note, I've noticed that when I tweet something negative about Klout, my Klout score drops within a day, even when the frequency of tweets has not changed. Coincidence -- I think not.
Meanwhile, five years after Gillin's BtoB column about the need for metrics, we still need reliable metrics. Klout is interesting tool, but it is hardly comprehensive or accurate. I also check out TwitterGrader.com from HubSpot. I find that a more useful tool because I have not seen the same inaccuracies that I've seen with Klout.
Let me know if you're experiences with Klout and TwitterGrader are different.