Yesterday’s post proved a few important things to me. One, when someone like Chris Brogan re-tweets you it can drive a lot of thoughtful activity on your site. Two, because of #1, if you think that a conversation might get started don’t put up the post / tweet about it and then head off for a 4-hour dinner – I imagine the activity would have been that much greater if the comments were going “live” immediately, rather than waiting for my approval (most came in a very short time frame). And three, perhaps most importantly, I might be onto something interesting here.
So the original question was whether Gladwell / Tipping Point theories, particularly in relation to mavens, connectors, and salesmen, apply to Twitter users. As a starting point, I looked at follower / following ratios as a distinguishing characteristic. The most interesting comment for me came from Amanda – wondering if I looked at re-tweets and @ responses. I hadn’t, and I should of, because it looks like it could add a lot of richness to the analysis.
This is where I start brainstorming out loud. Different types of “tweets” are used to different types of things. While some are more easily measurable than others, we could probably break them down into a few categories. Re-tweets (RT’s) are one easy one, and @ replies are another – both easily measurable. Other possibilities include tweets that link back to your own posts, and tweets that link to other posts you have discovered. For the sake of simplicity, lets ignore all others and start with those.
So an RT makes what would seem to be a clear statement – someone else has said / linked to something that I thought valuable enough to promote to everyone else. This activity would certainly not fit into the maven category. If forced to pick, I think it fits best in the connector group – people that link us up with the world.
An @ reply is something that seems very different. In general, someone has said / asked something that you think you should respond to directly. I feel this activity would certainly not fit into the connector category… and I’m thinking it best represents salesman – those working to make others want to agree with them.
Linking to your own blog posts seems straightforward to me – a Maven activity. You believe you have, or have created, some new information / context that people might be interested in.
Linking to other people’s blog posts… well that seems a lot like the RT thing. Someone else said something you like, thus you are trying to connect it with others whom might be interested.
So… this looks promising. Based on looking at a few different types of tweets, they seem to bucket fairly naturally into each of the three categories. Now as referenced in the previous post, I think this social media stuff is blurring a lot of lines here, but perhaps some benchmarks could be set to determine which category each person fits best into. In the simplest terms, if you RT a lot, you are probably functioning mostly as a connector; if you @ reply a lot , you are probably functioning most as a salesman; and if you link to your own posts a lot, you’re primarily functioning as a maven. What “a lot” is… well that’s open to debate.
Does that make any sense? And if so, anyone interested in collaborating on a project to figure out the numbers behind this thing? There might even be potential to cross-reference it with the follower/ following ratios… but that might just be confusing.
{ 11 comments… read them below or add one }
Denis, I like this.
I decided to check it out for my username, PurpleCar.
I went to search.twitter.com and searched on the term RT purplecar. You can find the results here:
http://search.twitter.com/search?q=RT+purplecar
I retweet many more times than I’ve been retweeted. (RT is shorthand for “retweet” and its the most commonly used syntax in my Twitterverse).
The results are pretty similiar if you use the full word “retweet” with my username. A lot are me asking for links to retweet:
http://search.twitter.com/search?q=Retweet+purplecar
I used Tweetscan.com next. I searched on the term “@” by my username. It isn’t a great search. The results aren’t very clean:
http://tweetscan.com/index.php?s=%40&u=purplecar&site=
I went back to search.twitter.com and just searched on my username:
http://search.twitter.com/search?q=purplecar
A lot of @’s to others, where I am usually engaging in mini-chats.
I didn’t bother looking up when I referenced my own blog because that, for me, is very rare. Out of all my 7,000+ updates I’d venture to say I’ve done that maybe a couple dozen times. With our current tools, it would also be difficult to ferret out which URLs I posted went back to purplecar.net. All of the URLs I post are shortened, which wipes out search-term potential. The rare times I look at wordpress stats for purplecar.net, the twitter referrers are from my bio or other people’s tweets. You’ll have to ask your readers if they know a better way to assess the ‘maven’ status.
Your theory fits for me so far. I retweet a lot, probably about as much as I just plain tweet (with no links, no @ replies, no RTs — to go further, you will definitely need to measure that “spew” factor. Let’s call them “plain tweets.”). A lot of plain tweets may indicate a maven who is positing questions (you can suss that out by looking at the amount of replies they get). A high number of plain tweets may also indicate a newbie or a blow-hard. That kind of judgment would have to be qualitatively made instead of quantitatively (like counting @replies).
Anyway, there is my own data for you to look at. Let me know if you think my own assessment of “connector” is good for me. Also, let me know if you find any other good search tools/parameters. I’ve discussed “authority” on twitter just recently (and ended up getting my picture on TechCrunch for it – LOL); that whole discussion led me to think about ways to measure value on Twitter and social media in general. A lot of the same ideas were discussed. Thanks. Let me know what you come up with.
Thank you – that is amazing feedback input. And once again, if you have me thinking more about this… particularly about re-tweeting. I’d focused on one’s own activity… but how often people RT you could be particularly valuable… perhaps for identifying “successful” mavens?
I’ll try to run some rough numbers across a few people and see what I come up with…
Funny, I was thinking something very similar. Although I was assigning Gladwell’s roles to separate sites. Example, I see Twitter as a Maven. Facebook and Linkedin as Connectors, etc.
I love what you’re doing. I teach smm @ ucla + would love to collaborate w you on figuring this out.
I’m out to dinner right now + will put more thought into this later. Please contact me -email or @beverlymacy on twitter
Actually most stats are already available on Twitter-friends.com.
Don’t we all adopt multiple roles? I promote my own (occasional) blog posts, I directly RT interesting tweets, and often post ‘what I’m reading’ if I found the article/whatever other than from a tweet. That’s when I’m not exchanging trivia about my daily life and responding to others doing the same – which I don’t believe fits anywhere!
Interesting stuff.
PeeeBeee, you’re right. Gladwell’s types are too confined to properly categorize all tweeters and all tweets, but it is interesting how much actually does correlate to his theories.
I suppose yours (and my!) use of Twitter for random thoughts would simply correlate to Gladwell’s mavens, connectors and salespeople living out their lives. Those types are doing those behaviors 24/7. I think we just want to categorize the MAJORITY of behavior on Twitter. Everyone is allowed some random thoughts, no doubt. We’re all human, after all.
You bring up a good point, though. We should qualify a percentage of tweets, perhaps 80% salesman behavior qualifies that tweeter as a salesman. The other 20% can be noise. What do you think the threshold would be? 75%? 60%? What percentage of a person’s tweets need to be solidly categorized within a Gladwell type to be safely considered part of that category?
Interesting question. Good point.
Meant to say above: “those types AREN’T doing those behaviors 24/7″
peeebeee / PurpleCar: I’m thinking you are absolutely right. My (evolving) hypothesis is that while one of the things about Twitter is that it enables us all to adopt multiple roles, that there are some metrics that can be applied to determine which role best suits are behavior.
As PurpleCar notes, it’s probably around thresholds – which I imagine can only be determined through quantitative analysis. The trickiest one, I think, will be “mavens”. My hunch is that the majority of activity will be around connecting and/ or “selling”, so the threshold for maven activity might be low.
However, putting together two pieces of data – in my mind- holds a lot of promise. The first is % of posts that direct readers back to one’s own blog post / idea. The second is direct response to such posts through Twitter – via RTs and @s.
But as I start playing with the actual numbers, I realize how tough this may get – particularly since @ replies are used in a variety of ways, and are often in conjunction with an RT.
Leads me to think a lit more qualitative analysis may be required than I thought – judgement calls about whether responses are being used to connect one person’s idea with another individual, or to provide some sort of answer to a question.
Hmmm…
Qualitative analysis is not to be feared. One can design a point system. When I was a grad student, I was tasked with devising a qualitative point system for scoring math strategies used by elementary school children. The data were extremely varied, as you can imagine. I made a 10-point scale using different characteristics, like horizontal or vertical orientation, cross-outs, errors, etc. I scored each sheet using the scale and I ended up with a whole pile of quantitative data.
Tweets of all kinds can be similarly measured. Yes, some Tweets will have to be judged down one side or the other. If I found a math problem that was written diagonally, I had to decide if it was more horizontally or more vertically done. Those decisions were mine alone, but I could have easily formed a committee to make them instead.
So it can be done. But who would spend the time and money, though? Devising a scale, translating the qualitative data into a mass of quantitative data, then providing an analysis of the results would be a major undertaking that would yield a margin of error as big as Manhattan. In other words, the data aren’t worth it. Twitter, and any online conversations, should be regarded as trends, fads, not law. Trends need to be assessed quickly and simply. I’m sticking with my time-span keyword cloud proposal for this.
I’ve been running an experiment for the last 4 weeks on twitter. Would love to share my thoughts and findings with you offline. They align really well with some of the things you are saying.
{ 5 trackbacks }