I’ve managed to cause quite a fuss after my previous post about the main players in the EU-Twittersphere. So much so that some more reflection is needed, something that, due to the lack of nuance of 140 characters, I’ve been thus far unable to provide to (among others) @mathewlowry and @FabienCazenave.

I think the essence here is that TweetLevel, the tool I used for the main players post, tries to take something rather complex – how good someone is at using Twitter – and to make it incredibly simple, to boil all the complexity down to one number. After all Edelman are in the PR business, selling social media strategies to clients, many of whom probably have little clue about the technology. Making it simple is the essence of their business. It’s not that far from what I do professionally – trying to get politicians and campaigns working online.

So is TweetLevel actually any good? I essentially think it’s intuitively about right, and that’s what makes it useful.

Take @AlexAlvaro for example – he’s a German FDP MEP who basically uses Twitter to broadcast to people, and follows just 19 people. That’s not a good strategy – Twitter is about interaction and conversation, and his score is a measly 31. @JeanineHennis, Alvaro’s Liberal colleague from the Dutch VVD scores 60 on TweetLevel – she actively replies to @replies sent to her, asks for opinions, and genuinely interacts with others via Twitter. Her strategy is better and she scores more in the system.

All of this is intuitively right – I appreciate Jeanine’s strategy more than Alex’s, her approach is the one I would advocate to clients, and now there’s a handy little tool I can use to lend a little more credence to my assertions. As the TweetLevel About Page shows, there is some decent methodology behind their site, and I think that’s reflected in the outcomes, and users are also given some practical tips on how to improve.

So TweetLevel perhaps tells us what we already more or less know or feel, but that doesn’t stop it being useful.


  1. I have risen by 3 points within 10 days, so I am 2.5% more influential now…?!

    Don’t think I’ve made much of a difference in that 1 1/2 weeks, so what does the change tell us?

  2. It’s not hard to work out why @bueti scores highly – if you @reply to him he actually replies back. As does Jeanine Hennis. Alvaro (generally) doesn’t. So old politician who is actually open scores more highly. Not that complex.

  3. Bente, that’s some pretty cool analysis you have linked to, thanks!

    I am not sure that the question is young meta-power vs established parliamentarian power – take the Vice-Chair of the GGEP (Greens), Reinhard Bütikofer, for example, who scores a respectable 48.5 on TweetLevel.


    And yes, you are right when saying that «the methology is based on a certain idea of “ideal” Twitter use» – that’s exactly the problem. It’s what social media network experts tell you the whole time…

  4. I forgot to post this article that is relevant for this discussion: “New data on Twitter’s users and engagement”, http://themetricsystem.rjmetrics.com/2010/01/26/new-data-on-twitters-users-and-engagement/

  5. You’ve touched upon a complicated issue, Jon. TweetLevel is just one of several tools out there that tries to measure something as untangible as “Twitter influence”, and I’m not sure how well their tool is. I looked at their methodology, and to a certain degree it makes some sense: http://tweetlevel.edelman.com/about

    But the methology is based on a certain idea of “ideal” Twitter use (you get a higher influence score if you retweet, if you follow people that follow you, if you interact (Edelman calls it name pointing) on Twitter, have lots of updates, etc).
    This is NOT a tool that measures quality of content, but your Twitter interaction.
    (I ended up with 53 and they kindly recommended me to get more people to retweet my stuff:-)
    I wonder if Edelman measures retweets in a similar way that Google measures influental links – if so-called “influental” Twitter people retweet you or talk with you on Twitter, will that give you a higher ranking? If we think about Google and their algorithm (from reading The Search by John Batelle), some links are more important than others, and if influental websites (based on ex traffic) are linking to your site, your site gets higher search ranking.

    When it come to Twitter influence and politics, this tool illustrates a fascinating paradox. Young, active folks on Twitter (like all those you mentioned in your previous post, Jon), can be more influental in this medium then well-established MEPs…
    Meta power (Twitter influence) vs. parliamentarian power?

  6. Let me be a little less diplomatic than Julien: I think, quite honestly, the tool is crap.

    I provide one illustrating example and rest my case: the Swedish presidency web team @se2009eu has a trust-level of 16.1, and the funny fellas of edelman say about that: “Your trust score is low because the content that you tweet is either not credible, interesting or newsworthy.”


    With credible, interesting or newsworthy they mean – exclusively – how often your tweets earn retweets or references. What good is that? If everyone used Twitter with the ambition to be retweeted and referenced, we would soon be overwhelmed by retweets and /via and RT.

    Well, maybe the tool is not so useless after all – if I could look for people with low levels of trust, I might be able to find some really unique and original users 🙂

  7. I agree with Julien : tweetlevel tells you if somebody is good at tweeting-retweeting-following-being followed… Nothing about the content, relevancy or even truth of the tweets.

    Though, it is indeed not completely wrong to compare people tweeting about the same topic. Not so wrong neither on your idea of comparing people.

    But we don’t really know by how much score difference is there a true significance: what’s the difference between a 48 and a 52 ?… And there can be people who tweet a lot and retweet less, as other will tweet very few messages but get heavily retweeted.

    I believe that a good indicator to watch on the influence of a person is how many time this person got #followfriday-ed Though again, not every body post #ff tweets…

    Complex issue…

  8. Hmmm, well, on reflection I should have spent more time on that post!

    I personally have rather little incentive to interact with either Alvaro or Hennis – they are from a political group that does not massively concern me. But what does concern me is how politicians use the web, and, conversely how the web changes politicians. Some (like Alvaro) use the web as a 95% one way street to broadcast. Others (like Hennis) use it as a way to build some genuine discourse, and as a matter of general principle of using the web as a politician I prefer the latter. It’s not the typical ‘I am a smug politician so listen to me’.

    Compare like with like, make lists and groups, and TweetLevel gives a reasonable indication of who is behaving like a human being, and who is behaving like a broadcaster.

  9. I was referring to you last post, where you mix up all of us, even though you kind of separated the lists.

    But take for example Hennis and Alvaro: They are both in ALDE – maybe their strategies are absolutely complementary, maybe the communication behaviour as it is fits perfectly into the goals of ALDE. It’s nicer for us because we can interact more easily with Hennis, but on the other hand Hennis is writing a lot in Dutch that I don’t understand and ignore when it appears in my stream while a line by Alvaro makes me aware of things I might not have noticed in Hennis’ tweets.

    What I want to say is quite simple: The Edelman tool is just helpful if you have already understood who is who, who interacts with whom and who has what kind of background. To interpret the numbers, you have to put them in a comparative perspective, and to be able to do so you need good knowledge about all this.

    So the numbers are not helping any outsider, they are just simplifying the process of selling this pre-existing in-depth knowledge to people who are easily impressed by numbers from 0-100. Oh yes, and you can also use scores for self-marketing purposes afterwards, telling how important your clients are now after you advised them…

  10. Yes, and that’s precisely the reason why I’ve tried to compare like with like! So between me, you and @kattebel it makes sense. Between Alvaro and Hennis it makes sense. But between Quatremer and Hennis it does not.

    So I take their tool, add some analysis, and make something useful of it…

  11. The problem is that you still have to see these numbers in comparison with others – without you making a list of people you think belong to the Eurotwittersphere and entering their names into the generator and putting them up, these numbers don’t say anything.

    In addition, these numbers are telling you something about quantities and not so much about qualities. They don’t tell what you should be talking about or what kind of tweets are retweeted or what kind of things you retweet. In the end it is clear: Those who interact more write more and get more replies or RTs, just because they communicate – but whether this is any helpful for you or for anyone else remains unclear.

    The mix of these figures also doesn’t tell you about the strategies you as a Twitter user need to apply to get your messages through.

    Maybe someone like Alvaro thinks that it is enough to use Twitter as a one-way channel, maybe this is absolutely effective for him? Is Jeanine Hennis really twice as influential as Alvaro just because she interacts more? Should it be a goal to be like Hennis – and if yes, why?

    These kind of lists make it easier for communication people to sell their advice, because they are simple. That is all.

    But there is actually not much use in comparing my score with the score of an MEP or with that of Journalist @quatremer – the communication we are doing on Twitter and the reasons and goals to be there are totally different and thus hardly comparable.

Leave a Comment

Your email address will not be published. Required fields are marked *