Social Media Week London: the Twitter Influence Leaderboard
* Featured, Social Media, Technology — By Ally Manock on February 4, 2011 5:44 pm‘Influence’ in social media terms was discussed a great deal in 2010 and seems set to continue to be debated in 2011.
As Brian Solis says, “unlike so many terms in social media, influence is not a new word invented or reinvented to suit the times. Its origin is Latin, “influere” which translates into something very interesting for the social era, “to flow into.” Almost everything we discover and share in social networks these days is done through our streams and to loosely translate the root of influence symbolises the ability to flow into the streams of others”
This idea does makes sense, however it doesn’t quite define ‘influence’ properly for me. There are lots of people ‘flowing into my stream’ (is it just me, or does that sound naughty?!) that are saying a lot, but meaning very little. They put out a lot of messages, but I don’t care about them (note to self: do a cull of the people I’m following on Twitter).
Wikipedia says that “social influence occurs when an individual’s thoughts, feelings or actions are affected by other people.”
This surely means that influence (on Twitter, at least) is actually impossible to calculate?
If person A on Twitter puts a tweet out saying that they had lasagne for dinner and it made two of person A’s followers hungry for lasagne….how would we know? Maybe only if they replied saying as such.
Who’s influential right now? Take a look at our Social Media Week Twitter influence leaderboard.
What is twitter influence?
What’s currently out there trying to calculate influence?
There are two main tools: Klout and Peer Index. Klout claims to be ‘the standard for influence’ and Peer Index claims to help you ‘understand your social capital’.
Both tools calculate influence slightly differently and give you some interesting data. But what they both really miss is the context. For instance, Klout says that Leonardo Di Caprio is very influential, but I don’t think he’s ever changed the way I think about something or made me buy something.
There are various discussions around these tools, particulary around Klout, such as this piece from Matt Owen at Econsultancy.
How we calculated ‘influence’ for Social Media Week
With all of the above still in mind, we thought that the upcoming Social Media Week in London gave us an ideal opportunity to have a little bit of fun and see if we could have a go at tracking the influence of people tweeting with the hashtag #SMWLDN.
We have developed a ‘Social Media Week Twitter influence leaderboard’ which will track in real time the most talked about subjects, people and events of the week.
We’re going to be very open about our thought process and how we have calculated influence. This is because, of course, we don’t claim to have answered the question. This isn’t another Klout or Peer Index.
We’d like this to be a bit of fun for everyone (we know there are some competitive social media peeps out there who will try and tweet their way to the top! Ok…maybe I’m just talking about me…) and we’d like the Social Media Week Twitter Influence Leaderboard to add to the discussion around influence.
The numbers behind the Social Media Week Twitter influence leaderboard
We had a think about the metrics that are available to us on Twitter to see if we could determine who is the most influential person on Twitter that is tweeting with the #SMWLDN hashtag.
No. of followers. Yes, it’s a metric, but as we have said, it’s not the only one. Popularity doesn’t mean Influence.
No. of times the person has tweeted to the #SMWLDN hashtag. OK, that helps to show us how active they are being on the Twitter hashtag for Social Media Week.
No. of times those tweets have been retweeted. Yes, the equivalent of the Facebook ‘like’. A thumbs up for that particular message.
…but it would also be good to see what the reach of those retweets is, so we looked at the No. of followers that the retweeters have.
This left us with tweets that mention the original tweeter, but that isn’t necessarily a retweet (e.g. “Wow, @ally_manock ’s tweets today are fantastic. Plus, I hear she’s a lovely person too.”). Ahem. So, we have also looked at No. of mentions of the original tweeter.
Obviously, the one thing we can’t really measure is the holy grail. What was the outcome? The action? Twitter is a mash of comments, thoughts, ‘likes’ and ideas. It’s what has been actioned/affected by those words that is the true measure of influence.
What do you think is the true measure of influence on Twitter?
We’re looking forward to the panel debate around influence at Social Media Week on Thursday 10th February.
Azeem Azhar, CEO and Founder of Peer Index is on the panel and we’re expecting some lively discussion! It will be interesting to see if the tweets coming out of that session affect our leaderboard significantly.
As we said, we’d like this leaderboard to add to the debate about influence. Do you think it shows useful data? Are we missing something?
Image credit: http://www.flickr.com/photos/bondidwhat/218782908/in/photostream/
Tags: Brian Solis, influence, Klout, London, Peer Index, SMWLDN, social influence, social media week, twitter, twitter influence
14 Comments
Influence is in the eye of the beholder! I tune into different people for different reasons. Perhaps I know them personally and know we like similar stuff, I might enjoy the content they publish or it’s because I know they’re knowledgeable about a subject and so present a broad and balanced perspective. When I look at the influencers for clients the numbers are just the starting point – that’s my long list. After that it’s about reading,watching,engaging and understanding them (and who they communicate with) before getting to the shortlist. It takes time and I can see why there’s a temptation to automate this process but for now I don’t think you can. Nothing replaces the human element and if identifying influencers is really important to you then take the time to really understand who they are and what interests them.
I agree. An automated tool can never do the human element, it can only be a guide.
Incidentally, there are two great blog posts about both Klout and Peer Index here:
http://www.wewillraakyou.com/2010/12/klout-is-broken/
http://www.wewillraakyou.com/2010/12/peerindex-twitter-spam/
Being shouty is a world away from being ‘clouty’.
Quantitative/algortim based techniques may overstate the former; to get to the latter reults may need to be weighted by a person?
I think your influencer machine is broken. If the organiser’s twitter handle @smwldn is at no.1 this shows there is no real human involvement. If I gave this list to a client and their own handle was no. 1 they would throw me out of the room.
You need to filter influencer lists by humans also.
Hi Andrew,
Good point. Thanks for adding to the debate!
We agree that the tool has its limits and that the human factor is important when it comes to determining ‘influence’.
Far from being a definitive measure, we wanted the leaderboard to add to the influence debate and provide a talking point for everyone involved in Social Media Week, and we hope that’s what it’s doing.
Hi Andrew – you’re completely right and we really welcome comments like this…it’s a debate that we wanted to start.
I wonder if we added ‘influence by hour or day’ if @SMWLDN would still be top? Something definitely to try for next time!
Totally agree, there are tools look at every kind of metric online, but without having the human element there to judge the person it will only be based on a calculation. I think the most valuable tool to measure influence is context, when where and what was said makes it more influential. I think measuring influence can be tracked live, but only properly measured once the event/moment has taken place.
Hi Guys
Thought this was an interesting debate – and indeed agree that what clients actually want to see is money in the till from their efforts in this area. However, PR teams have managed to build a successful industry based around a weighting system which is applied to the mentions they achieve – and this is fairly well commoditised: SCC coverage for each mention which varies by title, and then a weighting given as to the +/- tone of the piece. It is simple and yes – it does not give an absolute “bums on seats” figure – but what it does show is the “potential to buy” value of their coverage – or in your case the “how likely to buy” from the influence. J xx
I think your metrics are spot on. But then I may be biased…
This is a really interesting debate and one I will continue to follow as the company I work for, B2B Marketing, is about to launch our own online community and we are running a competition on the most influential community member during the first month. I will take your tips and thoughts above into consideration as we judge the winner…keep them coming.
Hiya – interesting metrics and good to see the results. I agree with the comments above this being just one element in measuring influence, as it’s not just about reach or ability to spread, but – putting my direct marketing hat on – it’s about outcomes.
I’m no statistician, but one weakness I spotted in the algorithm is that if you run several Twitter accounts with a reasonable number of followers, you can potentially game the system by continually re-tweeting one accounts tweets to the followers of the other accounts.
Unless there’s some kind of frequency cap, this’d get you to the top of the rankings. Although, again, I’m definitely not statistician
Absolutely Sam – great reply! I think you could easily game the tool as it stands at the moment.
It’s a little like Google (did I really just compare us to Google?!) in that you produce something, let it run, then figure out where people can spam the results.
We need to look into how we can prevent this from happening.
This is an interesting area. As I said at the panel session last week, I’m not convinced that what we’re actually measuring is influence so much as reputation and perhaps interest. Influence, as you note, involves changing behaviour. And let’s face it, even if people do express an interest in what others are advocating on twitter, there’s no evidence whatsoever that behaviour actually changes.
It comes back to the fact that there’s insufficient correlation between brand awareness and sales conversions. There’s plenty of evidence to support a correlation between brand awareness and intention to purchase, but it’s wrong to assume intention to purchase is going to result in 100% sales conversion. Perhaps even more importantly, there is some evidence to suggest that people present either an idealised or even blatantly fictional version of themselves in their online personas, so what they say they are interested in, or might be doing, may in fact bear little if any relationship to what they actually are doing/consuming.
So while the metrics you’re using are fine for identifying interest and reputation, I’m not sure they are really identifying influence. And even when it comes to reputation and interest, there’s always the problem with what I think of (perhaps rather floridly) as an “echo chamber of love”. Don’t get me wrong. I love positive reinforcement and I’m more than happy to acknowledge the great work of people and work I admire. But I’m also conscious that we’re all kind of celebrities in our own lunchtimes in the social media world. We love each other and we promote each others’ work at least in part because the act of cross promotion ultimately increases the profile of the social media domain. This doesn’t mean we’re being false or don’t believe what we say, but rather that we recognise that act of valuing each other is going to improve our value in the network, as well as the value of the network itself. This then produces an echo chamber of positive reinforcement, and can skew results of research looking at reputation and interest in a way that’s hard to counter.
But look, what I think is actually going on is less important than what business wants to know about social media. Business has been excited and amazed by what’s happened in Egypt and Tunisia and it’s just beginning to see how social can be used as a way of producing change. Business think of social media like twitter as being influence channels. So to a very large extent, your metrics are probably what business want to see. All metrics, as you say, are flawed. But it’s better to at least try to measure what business describes as ‘influence’ than to muck around with semantics (as I am wont to do!). And maybe through these kinds of discussions we can continue to hone what we measure, and make more reliable predictions about consumer behaviour based on what we observe?
Thanks for your great work and for the opportunity to contribute here. (And this isn’t an echo chamber. I really do mean that!)
Influence is multi-dimensional, you could try and gauge how influential someone is (1 being not 7 being very), you could look at influence in terms of what the recipient is likely to do i.e. share, purchase, enquire, ignore etc., where someone is influencing both in digital and ‘real-world’ environments, as already mentioned in this blog post – how they effect the recipient i.e. do they really recall and engage with the product.
Interestingly at the TFM&A conference this week a lot of speakers happily chatted about influencers throwing the word around without much consideration but knowing that it is somehow important. Emotional clusters are not new but understanding the impact of those cluster groups on their immediate network is a new way of thinking and how to leverage them even more difficult. It is one thing getting a software blogger to review your products it is another thing getting Mrs. Smith down the road to actually buy it.
I know I am not coming to any beautifully succinct conclusion here, mainly because there isn’t one yet, it is a work in progress. Tools like Facebook connect, continued uptake of social media on mobile platforms and more integration of auto-twitter, FB and share links on sites are steps toward ‘continuous’ connectivity (cybernetics theory) that would allow tracking and therefore measurement of influence. However there will always be gaps because at the end of the day you are dealing with humans and maybe it isn’t a good thing to measure absolutely everything?