Klout: about as much Science as anti-vaxxers
Ironically requiring the spelling to be correct in order to work with the meme.
The problem with online influence
The biggest problem with all of this rhetoric about online influence, whether about bloggers, Twits, or Facebookers, is the simple fact that more [email protected]/’Klout’ does not influential make.
It’s somewhat like saying that a business is successful because they get all the likes. Quite frankly, Likes don’t mean purchases, registrations, leads, whatever. They might be a precursor to, but it’s all meaningless drivel unless people actually take the action / meet that objective outcome.
Similarly, any claims about ROI through ‘brand equity’ or ‘brand awareness’ that doesn’t involve independent, evidence-based, large-scale market research is also full of rubbish. So it’s not just Klout or anything else. ROI is complex, not ‘simple to measure’ (for most brands, eCommerce it’s a lot easier than others) and involves actual math and statistics, not a magical SaaS tool you can get a 14 day free trial of.
The amount of salt one should take any ROI claims with.
OK so Klout is one of the topics that I really love to hate on. But given I haven’t hated on it on here yet, now is a perfect time.
So most people know what a Klout score is. It’s a magical number, and bigger numbers are always better. 0-100, a ranking of how cool you are online (is the word ‘cool’ still cool?).
But Klout’s primary problem: zero transparency, zero science.
Klout tries to explain its algorithm here. And as usual, you get some magical ‘statistiks’:
The majority of the signals used to calculate the Klout Score are derived from combinations of attributes, such as the ratio of reactions you generate compared to the amount of content you share.
For example, generating 100 retweets from 10 tweets will contribute more to your Score than generating 100 retweets from 1,000 tweets. We also consider factors such as how selective the people who interact with your content are. The more a person likes and retweets in a given day, the less each of those individual interactions contributes to another person’s Score. Additionally, we value the engagement you drive from unique individuals. One-hundred retweets from 100 different people contribute more to your Score than do 100 retweets from a single person.
OK so essentially, it’s a magic box that you need to make people RT and @reply you more. I’ve previously done a fun experiment and found you can arbitrarily pump it up with:
- Advertising, on Twitter or Facebook (duh)2. Posting cat memes or other content that is meaningless, but RTable — steal it from Reddit or BuzzFeed3. Just sending more content…
But the sad thing is that people seem to actually pay attention to this damn score.
I genuinely think that someone’s number of followers is a better measure of influence than Klout. Because at least that’s a well-defined, transparent number. Of course, it’s still bullshit to say ‘I’m influential because I have 5,000/10,000/100,000 followers’, but I’d genuinely pay more attention to someone saying that than ‘I have a high Klout score’.
My favourite article on this issue has to be this one by Sean Golliher about how he managed to reverse engineer Klout scoring to an R-square of 94% (i.e. pretty damn good). He found that it was pretty heavily associated with a logarithmic function of the number of followers and the number of RTs.
Now that’s hardly science or evidence based. I think I speak for everyone in saying that it doesn’t sound particularly ‘sciencey’. Not to mention the fact that it’s also based on the number of networks connected — so if you only look at your ‘Twitter Klout’ it’s likely less than your ‘Twitter+Facebook Klout’, even if you never use Facebook for work.
The problem with arbitrary scores
As soon as you start working on any kind of arbitrary score, or even a score calculated using an arbitrary number, is what happens if/when the score changes?
If you’re tracking Klout as a measure of success, what happens when they tweak the algorithm and your score suddenly shoots up (or down)? Have you become more influential? Or less? Or neither?
For marketers looking at analytics, we use time series data…that is, how something changes as time progresses. And if the metric you’re tracking is changing in nature over time, it makes the historical data effectively useless — one day you might be measuring apple ratios and the next orange ratios.
It’s a consistent problem — even look at the PTAT% metric on Facebook (% of ‘people talking about this’ compared to total number of Likes) which is a rudimentary measure of engagement. What happens when Zuckerbook changes how they calculate PTAT? The simple answer is: invalidation of any previous data. You have to start collecting from time zero once more.
So what’s the solution?
I hate being one of those ranty people who don’t provide solutions. Sometimes it’s hard to provide a solution because some things are complex. And ‘online influence’ is one of them.
But I think that really the ‘solution’ in my mind is that we need to stop looking at ‘social media ROI’ in terms of these arbitrary metrics, and start looking at actual ROI.
ROI is linked to a business outcome: engagement, followers, fans, Klout etc is not. It may be an indicator, sure. But spend too long looking at an indicator and you start treating it less like an indicator and more like a metric.