Improving the Klout algorithm



I, like many, have a love/hate relationship with Klout. It is an awesome idea and I agree with those that say companies want/need to identify social media influencers. The problem that Klout has is identifying actions in the virtual world that can drive value to a client company’s bottom line “in real life.” At least, the tech talking heads think this is the problem that Klout needs to solve to make money.

The algorithm that is supposed to solve this problem has a long way to go, at least when it comes to Twitter, and as has been documented time and time again, it can be gamed quite easily for those willing to put in the time and effort.

Even without gaming, you can raise your score just be tweeting a lot. And I do mean A LOT.

In itself, this is not necessarily a bad thing. A generic question like, “What are you up to tonight?” Something witty about a hot topic. A quote from a famous author. Twitter, after all, is meant to be social. Kind of like a hangout. These sweet nothings get some replies and retweets and will raise your Klout score in spades when executed in volume.

But will those Klout-raising messages drive ‘real life’ value for a company looking at a potential influencer? Probably (maybe?) not. And yet Klout will raise your score, leading a company to think that they can use your influence to help drive value.

So what are the actions other than replies, retweets and favorites that Klout can identify that can help companies?

Idea 1) Drive followers – as far as I know, Twitter charges for advertising, in part, due to the number of followers gained from a ‘promoted tweet.’ The value of a follower is up for debate but 10,000 followers is likely more valuable than 100 (assuming quality followers in both scenarios).

Earlier this month, Firestone Walker Brewmaster, Matt Brynildson, joined Twitter. When I spotted him, he was following a dozen people and followed by six people. I sent out a tweet urging people to follow him:

And what happened? Here is a sample:

He had gained something like 15 followers a minute later and gained over 120 followers by the end of the day.

The Klout result? My score did not change from this.

Idea 2) Drive activity around a hashtag – a hashtag can raise awareness of a campaign or event.

This is tough because you also have hashtags like #WhatYouShouldKnowAboutMe that aren’t tied to either of those. Nevertheless, the power to drive activity around a hashtag is pretty powerful.

Earlier this week, I declared #fcbs as the hashtag that people should use for the Canadian Breakfast Stout release this past Saturday. I wasn’t the first to ever tweet that hashtag but after my tweet on the 29th, over 135 tweets came in with that hashtag.

The Klout result? My score actually went down a bit, not that it should have changed much from only that many tweets. Some hashtags obviously trend to the point that thousands and thousands of users engage with them. But still, it went down?

Again, Klout is a powerful idea and the algorithm will obviously improve. Here are a couple suggestions for Joe Fernandez and Co. over at Klout:

1) Identify and reward driving behavior outside of retweets, replies and favorites.

2) Eliminate or lower the weight of tweets that include famous quotes and generic questions like “What are you doing right now?” Hell, maybe lower the weight of questions in general. Maybe ones that include “help me” or “how much” for instance. If I get a ton of replies to a question like that, it may mean that people are just willing to help me out but how much more “influential” does that make me?

Identifying those behaviors is probably something that Klout is working. Hopefully, they figure it out.