Thursday, August 26, 2010

The Power of One

One may not seem like a powerful number... but it is. Like the well-known Pareto Principle (or 80/20 rule), there is another semi-universally accepted theorem about the ratio of content creators to commentors to lurkers. This is the 90/10/1 or 1% rule. Author and friend, Jackie Huba, has talked about this rule a number of times, mainly in the context of what companies and organizations can expect when it comes to expected contribution rates in social networking activities and online community.

What got me thinking about the power of one the other day was a graphic created by eMarketer that appeared in a post titled, Why should brands bother with twitter? This table says why. I've embedded the table below for ease of reference but one can pretty quickly see that when it comes to engaging in social activities, people that use Twitter over index off the chart.


Upon tweeting about the "Why should brands bother" post, my friend, Liz Phillips, chimed in...


Wow, could this be true? Obviously there is some self-selection because people who tend to be hyper-social also tend to be on Twitter. But looking at the chart, the difference in "social activities" between Twitter users and ordinary internet users is 100-400% greater. Jackie Huba (who I mentioned earlier as being a pro on this front) clarified that most of the research around the 90/10/1 ratio has been focused on online community which is important to keep in mind. But might this number be shifting?

In thinking about the numbers and what looks like a dramatic shift in the 90/10/1 rule (which is different for social media than online community), one explanation for this might be what I'll call "the Facebook" effect. With 50% of Facebook users (now over 500 million) logging on to the service at least once a day for at least an hour a day, the general internet population is getting more comfortable and familiar with the idea of posting updates, links, pictures and video while also commenting and "liking" friends and colleagues content.

Getting back to Twitter, one thing that has always bothered me, especially when it comes to discussions around following to follower ratio is how many of the people I'm connected with actually tweet. I follow over 12,000 people and you might be surprised to know that I receive no more than 10-20 direct messages (DM's) a day and those are mainly from close friends and contacts. I also suspect (although I have no direct evidence) that out of the 12,000 people I follow, somewhere south of 20% constitute 80% of the updates (there's that 80/20 rule again). With 14,000 plus people following me on Twitter, I've also wondered how many people are actively monitoring my tweet stream. My guess is maybe somewhere around 1% at any given time.

To validate this number, I took a very unscientific approach. Last night while sitting in traffic (yes Oprah, I was at a complete stop when I tweeted this), I put out this tweet...


And then I waited... for 12 hours in fact before putting out another tweet. At least 50% of my responses came back in the first hour with the remaining 50% trickling in over the next 11 hours. The number of respondents was 78 (two coming via DM). As a funny side note, many people followed the rules and gave me a straight "yes" but several people got creative and a few even included one of my favorite lines, "I see you" in their response. Doing a little back of the envelope math, 78 responses is a little more than .5% of the number of people that follow me. Taking some statistical liberties, I'm going to estimate that the number of people who responded to me represented 50% of those that saw my initial tweet (the other 50% thinking, "why should I waste my time responding to you Strout.") Guess what, if you double that .5% [drumroll please]... yup, you get 1%.

What does this mean? To be honest, I'm not sure. Except that there is a power in numbers. Whether it is 80/20 or 90/10/1, it shows that behavior can be predictable. What I'm more interested in seeing is whether my theory about the Facebook Effect really has legs. If anyone has any research on this front, I'd love to hear about it. In the meantime, if you have a different theory OR would like to dispel any of my VERY unscientific research, please feel free to do so in the comments below.

I would love to send a HUGE thank you to my friends who (at least some of the time) pay attention to what I have to say. Most of you are already in my "Trust Tree" column on Tweetdeck ensuring that I am more likely to see your tweets versus those in my general "friend" stream. For those of you that weren't, consider yourself added (see, it does pay to play along sometimes). The folks that participated are:


@savvyauntie @slowrolling @GenuineJack @leslie @schneidermike @cbarger @smashadv @mtlb @jimstorer @ektrondave @FrankEliason @minesm @blogboy @joeyTWOwheels @CBMeitzner @angelicadevere @trishrc @jblock @Movember @adamgarone @ProPaintersAmer @kpitkin @willsmom @jpitkin @cselland @rehor @nealrap @glennbanton @techguerilla @KevinMSpence @TheChrisWalters @sdigregorio @Genuine @nickhuhn @marksylvester @micheliente @ecc1977 @banane @jenhorton @nciske @asilkin @RussThornton @michaelpearsun @mikepratt @asquillace @marting @mpace101 @AnkushKarnik @csoandy @JacquelinesLife @bryanstratton @stepchange @SiobhanDullea @vascocv @Aronado @mikeneumann @JohnKinATL @DrncPno @mkknief @Irdone @shawn @BigMikeInAustin @TTaxChristine @JohnKinATL @DrewCarls @Ed @gingerw @dead_tree_media @jtdoherty @myerman @CountXero @BlackDogPromo @dmcordell @DarrinSearancke @LevelTen_Colin @carriejbond @JeanTerranova @m750

12 comments:

  1. I was wondering what you were up to with that tweet. This is a very interesting experiment.

    I have to wonder, though, if the results would have been different if you'd tweeted your request earlier in the day. As I mentioned in my reply to your experiment request, I'm normally not watching Twitter at 6pm Central Time. I only happened to catch it as I was browsing around and saw your tweet in my "Met In Real Life" list's stream. I'll bet many of your followers had "gone dark" by that time.

    Still, it is quite fascinating - and possible proof to some ideas out there. Thanks for sharing.

    ReplyDelete
  2. When I first saw your tweet I thought you were testing your tweetstream or twitter account for some problem. Cool experiment.
    Thanks for sharing this, as well as all the cool stuff you share on a daily basis.

    ReplyDelete
  3. Interesting quickie test :) I wonder if any body skews the 1% rule on other platforms, or on Twitter for that matter, with their response level?

    ReplyDelete
  4. Aaron, that response rate might be higher than 1% -- according to my unscientific guess, of course. One thing to consider is how many of our followers are actually inactive accounts or bots? If you look at your Klout score (http://klout.com/aaronstrout), it shows your actual reach estimated at 5,000. (They have some way of filtering out inactives/bots.)

    If that figure is accurate, what does your response rate recalculate to? This is one of those things I'm interested in but like you don't have any solid research to back up my hunch.

    ReplyDelete
  5. I, for one, am disappointed that a certain @MelanieStrout didn't make the list! You guys crack me up with your seemingly 1:1 correlation of tweet activity.

    Speaking of correlations, my activity on twitter seems to happen mostly while waiting [traffic etc] or in idle moments like walking the dog. It's nice to know that there's always something interesting / entertaining to discover no matter when I happen to log in.

    Thanks for keeping it interesting!

    ReplyDelete
  6. @Elmer - I'm sure the time of day had an impact. Although part of me wanted to do it when things were a little quieter. To be honest, I was pretty thrilled to get 78 responses. I've also talked to a few other people that said that they did see my tweet but "knew that I knew that they keep a close eye on my tweets" thus semi-validating my assumption of the 50% rule.

    @V V - yeah, I wanted to walk the fine line of reeling people in and not skewing the test. Sometimes ambiguity is our friend. ;)

    @Leslie - interesting thought. maybe I'll give it a test on Facebook sometime in the future when people least expect it.

    @Connie - you make a fantastic point. I had thought about the spammer/fake followers but hadn't thought if it having that dramatic of an impact. What I will guess is that the number of "real" followers is probably somewhere between 5-10K as I know that there are a decent number of folks like my Mom that only have 15-20 followers and mostly use Twitter in passive mode. While they are technically "active," I don't think they show up on Klout [I'll need to ask @Gregarious about this one]. Back to your point, let's say the true number of "real" people following is half way, maybe 7,800 people, then my number of responses goes up to 1%. Keeping with my line of thinking, maybe the real response rate is 2% thus slightly skewing the "Power of One." Thanks for making me think more about this one.

    @Nick - ha ha. I think Melanie didn't respond because she was home making dinner and by the time I got home, I let her know about the test. Had she responded, she would have theoretically become an "outlier." ;)

    ReplyDelete
  7. Interesting post.. like Elmer I caught this in one of my lists long after your original tweet. I wonder about the timing too and how that skewed the results. In any event, cool idea/execution/results. Would love to see the respondents placed on map based on their geo-location.

    ReplyDelete
  8. Interesting analysis Aaron. I happen to believe the 90/9/1 is kind of a crap rule of thumb with regards to communities too. In my experience, it applies to very large consumer communities and to customer support communities but not for other types of communities. It is used far too often as a crutch but there can be much higher engagement levels in well managed communities IMHO :)

    ReplyDelete
  9. Of your followers, you have to keep in mind that a chunk will be spam. Do you auto follow those that follow you? If so, that messes your stats up.

    Good post, thanks for sharing.

    ReplyDelete
  10. @Jim - yes, timing ABSOLUTELY had an impact. But in some ways, I liked it better than doing it early in the day. I didn't want people necessarily seeing my tweet because other people they knew were responding to me. I also didn't want it to get subsumed by all the noise that can happen. With that said, I do plan to do the test again sometime in the future. And I may also give it a whirl on Facebook to see what happens there.

    @Rachel - "crap rule." I love it. And I couldn't agree more.

    @Ryan - yes, that's a good point. Connie also pointed that out. Interestingly, I've been looking through "Twit Cleaner" recently and there are definitely at least 1-2K followers that are semi-inactive accounts. I don't auto-follow back so I know that a few (but not many) of the folks that follow me are legit. Glad you liked the post.

    ReplyDelete
  11. Aaron,

    As Elmer and Jim pointed out, timing has a TON to do with engagement.
    As I teach in all my courses, if you can accurately gauge when your
    audience is listening (and this is near impossible unless you set up
    the type of social recording I feel everyone should have around them
    at all times), then you'll be vastly more successful at engaging than
    if you just shout into the abyss.

    I confess that I didn't see your post. I also confess that I miss
    about 94% of the updates my girlfriend puts online. This isn't because
    I'm not interested (cue #bromance music) or thoroughly smitten (photo
    of @clownface3 here), but that I use Twitter as do many others - when
    it fits my life and professional schedule.

    That said, you might see my 40 RTs at 4:09AM when I visit the toilet.
    Or you might see the dozens of @ messages during rush hour (sure, I
    stop to Tweet too).

    But you might not hear from me directly for days or weeks unless I
    happen upon your Tweet or other content during the times I'm paying
    attention.

    WHICH brings me to the real point of this blather...one type of
    content isn't enough. There needs to be a balanced mix of
    conversational and sharing techniques if you want to engage a large
    crowd.

    The tweet I sent out that got the most RTs and response lately was this.....

    Photo: Deep-fried Twinkie at the NY State Fair. #wwjce
    http://tumblr.com/x47gzklym

    Which means the people engaging with me are only interested in food or
    trailer-trash adventures. OR it means that I was lucky enough to share
    it with more people in that one moment than I would have been if I
    shared it later or earlier.

    What's the key to this whole business? Testing and a quality timepiece.

    Thanks for listening.

    Jeff

    ReplyDelete
  12. Nice post Aaron. Very interesting. I'm with you and Elmer, timing plays such a big part. In fact, I just blogged about timing emails tonight. I missed your tweets or I would have responded.

    I'm always fascinated by people who use Twitter strictly as a newsfeed but don't tweet themselves. Though sometimes I refrain from participating too, esp. when I feel like I'd just be saying something just to say something.

    So I guess it comes down to your definition of reach. Because you could have created awareness- people seeing your tweet without answering- which still counts for something. But engagement, of course, is deemed as that actualized goal of mobilizing your readers to action. Still. I think awareness should be considered too.

    Very curious. In closing, don't mess with Oprah. She'll find you.

    ReplyDelete