Finding Your Influence Is Just A Placebo. Edelman May Be Harvesting Your Data.

We all want to know who in the industry is an influencer. It seems that the first place people may look to find folks to reach out to these days would be on Twitter. But how can you find out how influential these people are? One you probably rely on looking at their conversations and things that they post. Before Twitter, the best way was to look at those blogs listed under specific tags on Google blog search or Technorati. Now with more conversation taking place and Twitter becoming mainstream, it’s becoming more flooded with people wanting to talk and thinking they’re the best topic commentator – some whom are well known and obvious, but others who aren’t. So how to separate influencer from wannabe?

That’s where public relations firm Edelman has come in. They’ve created a “nifty little service” called TweetLevel and what it supposedly does is allow you to measure the influence of any particular Twitter user. Just enter in their username and (barring any huge traffic push), you’re going to be able to generate an influence number. Very much like how HubSpot’s Twitter Grader is, the number is probably nothing more than a placebo. Yes, as a marketer, you might be more intrigued to use that number to evaluate the worthiness of someone as to whether you should do some blogger/twitter outreach to them, but for all intent and purposes, the numbers generated through these apps to help “discover” your social media “worthiness” is nothing more than a placebo. It almost ranks as much as the number of Twitter followers you have or the number of lists you are on. These forms of “status” in social media mean nothing…but I suppose it makes you feel good from a more selfish reason, right?

TweetLevel.comTo make you feel all warm and gushy inside, measures your importance on Twitter on a sale of 1 to 100 – obviously the higher your score, the better you are over everyone else…or at least as it relates to influence. But what are these metrics?

Influence – obviously this is an important one because it’ll show you how to rank compared to all your friends & shows you whether people listen to you when you say it.

Popularity – are people flocking to follow you?

Engagement – are you having a conversation with people in your community or are you simply pushing out content that people may or may not care about?

Trust – what makes you so believable and can you affect people’s buying behaviors?

So now that you know of the main factors in how to influence someone, what does this all mean? While I’d like to think that offers something more than an arbitrary number that might give you some slight bit of advice on how to improve your interactions, I’m weary to think that the numbers actually will mean anything – hence my reference it to being a placebo.

But I bet to Edelman, it’ll mean a whole lot more. Why? It’s because they could potentially be harvesting your information to build a giant database that they could filter and build upon to provide to their clients. It seems that the way to get people to “volunteer” their data would be to make it somewhat entertaining, almost like a game. With HubSpot’s Twitter Grader, it makes it seem that you’re in competition with everyone else to try and have the highest number. But deep down, the database is being built that the creators could potentially harvest and use to help build a better marketing & PR program. It’s smart. I mean, there’s nothing proprietary or private being lifted, just an analysis of people who are truly active on Twitter who would participate in this entertaining platform and provide their Twitter handles. Edelman could then just use that information and, depending on the needs of their client, could just go reference any list and say “let’s reach out to these 100 active Twitter users because they have an ‘influence score’ of over 80” or something to that effect.

Another approach could be that Edelman would want to be able to rank Twitter users by their so-called influence and bring them to an exclusive event or have them pitched a specific demo or if companies themselves tried to assess their own “influence” using, then reach out and do some potential business development.

I’m not saying that Edelman has created a really bad service here. On the contrary. I think that it’s a good internal system for them to use, but it’s also exploiting the naivety of those who think that status (e.g. number of followers, Twitter influence, number of lists, amount of retweets, etc.) means everything. But, while that may be more of an internal battle for the individual, on the whole, this system will aggregate a genuine list of active Twitter users who are passionate about the platform and also their industry and find out which ones are worth pursuing as influencers.

16 responses to “Finding Your Influence Is Just A Placebo. Edelman May Be Harvesting Your Data.”

  1. znmeb Avatar

    Oh, man, where to start: 😉

    1. There are at least half a dozen Twitter rankers out there. There's TweetLevel, TwitterGrader, Klout, Twitter Analyzer, Twinfluence, and my personal favorite, Twitalyzer. They all have different metrics and ranking algorithms, and different “features and benefits”.

    2. *All* of these tools take advantage of the fact that *any* public tweet is available via a publicly-defined simple application program interface (API). In other words, nobody has to enter anyone's name into a tool for the tool to be able to poll the Twitter database for any public data! The tools can just go out and grab data. I've got scripts that collect Twitter data running right now on the machine where I'm typing this.

    3. InfoChimps is collecting and attempting to sell Twitter data. See the discussion here:

    Twitter Data Dump: InfoChimps Puts 1B Connections Up for Sale

    4. Twitter's “privacy” policy is a joke – “Twitter privacy” is an oxymoron. I'm surprised that there hasn't been a huge outcry over this.

    5. Since public tweets are “free” for anyone to collect, they don't have much value on their own. The business value must be *created* – you can't “mine gold” from raw tweets.

    6. The kinds of analyses being done now are simple – they represent what's feasible with a moderate expenditure of coding time by typical developers using modest amounts of computer power. The “real thing” – natural language processing, large-scale social graph analysis, sentiment analysis, etc. – requires much more.

    First of all, you need some PhD-level talent just to develop the algorithms. “Conventional” natural language processing algorithms don't work because Twitter is multi-lingual and the “natural language” of tweets is evolving as the platform's use grows.

    Second, the magnitude of the datasets and complexity of the algorithms create requirements for *large* computational resources. The base algorithms involve processing large multi-dimensional arrays. Nobody is going to invest in the hardware and software to mine tweets if there isn't a compelling business value to the results. And I'm *extremely* skeptical about such value.

    Ed Borasky

    1. Ken Yeung Avatar

      Hi Ed –

      Thanks for your very thorough comments. Definitely enjoyed reading them.

      I think that you're right…there are dozens of these Twitter ranking tools out there and I didn't mean to exclude any, if that was your point. I just thought that since Edelman created this system, it was similar to another large marketing firm making another, which is where I got Hubspot's Twitter Grader.

      Definitely the open API system and whatever public data is out there is where you can gather data, but while you do gather it, how much is relevant to the user? If you have people interact and they “self-populate” your database through a game methodology like with TweetLevel, doesn't that give you a better sense of those who are (1) interested in learning their influence, (2) avid or habitual users of Twitter, (3) are social media type individuals/personas?

      As for InfoChimps collecting AND selling Twitter data, I'm unfortunately up to speed on that debate, but just know that I think that selling data by any third-party or selling data is just wrong. It should be done organically and with a specific methodology. Edelman & Hubspot may only just be using it to find the right people who they might seek to reach out to. But I couldn't say because I'm not privy to any information.

      Your skepticism towards finding a compelling business value is probably merited, but I think, with respect to TweetLevel, the “business value” here is that they're offering a light-hearted social media lesson on how individuals using Twitter could increase their influence – through hard work, dedication, sharing relevant information, conversating, etc versus offering “get-rich quick” schemes like making 90,000 followers in 60 days.

      Thanks very much for your comments.

      1. znmeb Avatar

        “If you have people interact and they 'self-populate' your database through a game methodology like with TweetLevel, doesn't that give you a better sense of those who are (1) interested in learning their influence, (2) avid or habitual users of Twitter, (3) are social media type individuals/personas?”

        I think the self-population only indicates (1) – (2) and (3) are really derived from the data, or should be. 😉 Another reason why most of the rankers only use self-population is computational cost – exploration of the social graphs and tweet text takes up database space, uses compute cycles, network bandwidth and Twitter API calls.

        “I think that selling data by any third-party or selling data is just wrong.” Well … I've actually talked to the InfoChimps people, though not recently. They collected a huge Twitter dataset a year ago, when the Twitter user base was much smaller than it is now. They posted it – for free – for a day or so. Twitter asked them to take it down until a contract could be negotiated and then never followed through.

        In any event, there has been quite a bit of dialogue since InfoChimps made their latest announcement. You can see most of it on the ReadWriteWeb posting and the links there. I never weighed in personally because

        1. I can't afford to buy a dataset from InfoChimps. so my opinions have little clout.
        2. Twitter can simply pull the plug on them, preferring instead to sell data to Microsoft and Google, who are negotiating in good faith with legal counsel, rather than simply collecting data and trying to resell them on the open market.
        3. I was wrapped up in one of my other hobbies, the openSUSE 11.2 Linux release. 😉

        “With respect to TweetLevel, the 'business value' here is that they're offering a light-hearted social media lesson on how individuals using Twitter could increase their influence – through hard work, dedication, sharing relevant information, conversating, etc versus offering 'get-rich quick' schemes like making 90,000 followers in 60 days.”

        That's precisely the motivation behind Twitalyzer, and in fact, the reason I like Twitalyzer. The Twitalyzer metrics and reports are actionable – you get recommendations from their analysis. Still, as Peter Holmes of Reason Partners says:

        “Sure, many think social media is free. But is it really?

        “The fundamentals of marketing still apply. And as much as the 'conversation' meme has made its rounds, don’t fall for it. Marketing is still persuasion whether it’s online or not. Perhaps a different, more engaging, creative and even outrageous type of persuasion. But persuasion all the same.”

        We really should invite Eric T. Peterson (@erictpeterson – Web Analytics Demystified – the creator of Twitalyzer) in on this discussion, because he really understands the whole process from end to end. My piece of the puzzle is “only” the Twitter API and the heavy matrix / tensor math in natural language processing.

        1. Ken Yeung Avatar

          Thanks for your perspective on the whole matter. I think that this has been an interesting discussion from what I originally wanted to call out. Definitely wasn't about the API or selling of data per se, but rather on how all the services created by these companies wasn't really going to provide the end user (the Twitter population) much information than a number that is wide open to interpretation.

          As for InfoChimps, definitely it would seem that Twitter would probably go after the big fish like Google or Microsoft than allow a contract with InfoChimps to go through. I think that there may be a potential for that, but again, if it involves harvesting my data via API or whatever is public and then selling that to another company for exploitation, then I think that just equals how people buy lists to target using emails but just more in the social media sense.

          Would love to get Eric's thought on the whole discussion here. I must admit that it's taken the whole conversation to another level that I'm a bit chagrined and definitely interested to see how it progresses. Your understanding from the API and the natural language processing has been pretty thought-provoking.

  2. Dharmesh Shah Avatar

    Great article and you make some exceptionally good points.

    Disclaimer: I'm the developer of Twitter Grader.

    Regarding the value of rankings and grades: Though all of these tools (including mine) are imperfect, I think they do bring value — for some people. I built Twitter Grader primarily for myself because I wanted to get a sense for what the power/influence was of people that were following me. I found that simply looking at the number of followers was not sufficient. The algorithm has gotten increasingly nuanced over time, and I personally think the Twitter Grade is a reasonably good reflection of what it is trying to measure. But, like spam detection systems, it's a constant battle with those that seek to manipulate the results.

    In terms of the business motivation for Twitter Grader, it's two things: One, it builds brand and drives potential customers to our business. Second, the data is fascinating, in aggregate and we periodically publish papers and blog articles based on it.

    1. Ken Yeung Avatar

      Hi Dharmesh,

      Thank you for your comments. I think you really make some good points yourself.

      However, I would like to challenge you on the fact that you said that you believe some of these tools, although imperfect, bring value to some people. If you're thinking end users like myself who have been to those sites to ascertain our value, then I would tend to disagree. BUT, if you're saying that the only people who gain value are the developers and companies of the software, then I think you're right.

      I get how you say that you wanted to get a better sense of what the influence of people were for those following you, but I'm not sure if people truly understand the algorithm. Do you believe that this is something that could be confusing? Just because some of the people whom I follow are deemed Twitter “elite”, does that make them any more influential than other people I follow – just because they have a higher “grade”? I think this is a trick question and I don't know the answer myself – because yes…folks like Brian Solis, Pete Cashmore & Robert Scoble could be way more influential than someone who interacts with others more occasionally than the other so-called “A list” ce-web-rities…but it can also be applied vice versa. But end users wouldn't know this because the “A list” folks have a higher grade thereby making it assumed that they're more influential.

      You're right though…it is a constant battle to fight those who seek to manipulate the results.

      But for your business motivation, since it's part of Hubspot, are you using Twitter Grader to amass the data to leverage it for your clients? Not necessarily sell it – which I hope you aren't, but perhaps saying that you have access to this data to which they won't see, but trust you'll use it to benefit their campaigns? For example, you execute a campaign and then internally flush out who you think would fit the mold from those who have participated on Twitter Grader and then generate a list to do some outreach or some other program with?

      What brand does this build to have people try and figure out their grade?

      Definitely appreciate your insights and hope to continue the conversation…

      1. Dharmesh Shah Avatar

        OK, I see your point. You're wondering why people would want to know their grade. That's a different point than whether the grade itself has any meaning/value.

        I'm more focused on the grade itself, and using it as a way to measure how social media activities are going, and finding the influencers. We use this data to help our customers already.

        We have no plans or desire to sell the data. Honestly, hadn't even really thought about that. The reason we make the grade available through the free tool is that it helps make the tool better (simply by folks telling me that a certain grade just doesn't make sense).

        Having said all that, it's all just my opinion.

        1. Ken Yeung Avatar

          Oh, I'd love to hear more about whether you think that the grade itself has any meaning/value. I don't think it matters whether people would want to know their grade because no matter what, from a inherent standpoint, we're all eager to find some way to valuate our activities. I'm always wondering what my “influence” is and I've gone to TweetLevel and also Twitter Grader to find out – and from a personal standpoint, I'm always eager to find out how to improve it, but I do know that these numbers don't mean much…

          Or do they have meaning? Should we get caught up on the fact that I have a 99.9 grade on Tweet Grader while folks like Brian Solis has a 99.99?

          I'm also definitely interested in hearing more about how you're using this data to help your customers – not from a “tell me your trade secret” type standard, but what types of things are you using from Tweet Grader to help your clients?

          1. Dharmesh Shah Avatar

            As the tool has grown in popularity, the percentile grade (99.9) is less useful than the actual rank, because so many people will have a 99+ — by definition about 1% of all users. The ranking provides a much more granular value.

            Twitter Grader looks at a number of things: followers, follower/following ratio, engagement (retweets), list inclusion (new) and presence/absence of spammy behavior.

  3. Jonny Bentwood Avatar

    Interesting post and following conversation.

    Just to let you know that I created TweetLevel and of course work for Edelman. I want to be very clear first of all that your premise that we created this was to harvest data which we can resell is absolutely false.

    As some of your commentators have correctly stated, all the data that we use if publicly available.

    Our primary reason for sharing this tool is to engage with the Twittersphere. We believe that engagement and trust are crucial components of online conversations – at Edelman we are proud to be part of this community and will continue to share our thinking and look to ways that we can understand influence within social media.

    At no stage do we believe that we have solved the 'holy grail' of defining twitter-influence but we believe our tool is a good step forward. The fact that there are so many different approaches illustrates that people are getting lost in the Twittersphere understanding who is important as oppose to merely popular. When the focus moves to the micro-topic level this becomes even more crucial as we get to identify specialist influencers in niche communities whereas before we would have been lost and merely resorted to naming the mega-stars.

    I think this comment sums up our raise d'être in itself. Your post has triggered a debate at which Edelman (and many others) are engaging.

    1. Ken Yeung Avatar

      Hi Jonny –

      Thanks for your comments and feedback.

      I would like to clarify that when I said “harvest”, I meant it in no malicious way like spamming or anything like that. I was alluding to the fact that while people were filling out this data, could Edelman or anyone else creating a service like TweetLevel be perhaps aggregating this data to help better service potential customers.

      I'm happy to hear that this tool was created to engage with the Twittersphere, but what, may I ask, is the reward for you creating this tool so people can figure out their influence? I'm very curious as to the meaning behind the service which I think is a pretty interesting tool.

  4. Jonny Bentwood Avatar

    Thanks for the clarification Ken

    The reward for Edelman creating TweetLevel is two-fold.

    Firstly it allows Edelman to engage in this discussion around influence, social media and trust. These concepts are at the heart of what Edelman stands for. Having tools such as this shows that we are serious about understanding the complexities of online engagement, promotes our brand and helps strengthen our foundations which perhaps gives us the right to talk with some authority about the issues which we hold dear.

    The second point relates to my earlier comment. At Edelman we can use TweetLevel independent of any user inputted data on a micro-topic level to identify specialist influencers in niche communities whereas before we would have been lost and merely resorted to naming the mega-stars.

    1. Ken Yeung Avatar

      Hi Jonny,

      I think you've answered my question…I was definitely wondering what “benefit” Edelman gets out of putting together a service like this besides it giving people an idea of what their influence is – although I believe that any value from this is merely a placebo. But in response to that, your second point is probably the most valuable – that you can use the user inputted data to “identify specialist influencers in niche communities”.

      So it seems that Edelman is not always interested in reaching out to the mega stars when pitching or running a campaign. Instead, it's more about being targeted and finding the right “true” influencers?

  5. Jonny Bentwood Avatar

    Ken, I think you misread my comment. To clarify we do *not* need any user inputted data to understand who the key people are for any given micro-topic as all the data is freely available on the web.

  6. philgo20 Avatar

    These tools are irrelevant ! Influence can only be seen as relevant to a domain, topic or area of ese xpertise. I don't think Shaq or Ashton Kutcher are very influential in programming circles…

    Just wrote a post about their lack of relevancy

  7. philgo20 Avatar

    These tools are irrelevant ! Influence can only be seen as relevant to a domain, topic or area of ese xpertise. I don't think Shaq or Ashton Kutcher are very influential in programming circles…

    Just wrote a post about their lack of relevancy

Leave a Reply

Discover more from Ken Yeung

Subscribe now to keep reading and get access to the full archive.

Continue reading