Turn Down the Volume

Watching Static

Image by xJasonRogersx via Flickr

Twitter is a great tool, when used effectively. But on my personal account, I’ve just had way too much “noise” lately. Basically, I’ve followed people that I wouldn’t really listen to. They just distract from those that I want to hear. Before you get all defensive, I’ve heard both sides of the argument: follow everyone you wish, and use tools (like Tweetdeck) to filter the noise. To me, this is simply one useless step that, in effect, is noise in your noise-removal. Under certain circumstances (in business-brand-building Twitter accounts, for example), you should follow a lot of people. But for my personal account (and yours), I’d suggest eliminating the noise. Here’s how.

Formulate a Goal and a Strategy

Goal: I knew I wanted to test Dunbar’s number, as explained by Malcolm Gladwell in the Tipping Point. So my goal for total number of people that I follow was 150. I had approximately 700 people I was following. Time to get rid of some junk.

Strategy: I knew there were tools out there to rid my account of people who hadn’t tweeted in 30 days, were known spam accounts, and fit other “junk” rules (sorry, Guy Kawasaki – you put too much noise in my signal). I would use those tools, and then whittle my account down by hand to get only the signal stream I wanted. Here’s how it went down, what tools I used, and some of my results.

The Tools I Used

Cleaning By Hand: The Criteria I Used

As soon as I finished the automatic cleansing, I went to manual vetting. But what criteria did I use? (great question, Kelly) Here are the 3 main criteria (the ARC) I used when going through the 300 or so remaining users:


This fundamental questions was “do they tweet?”. This question is one of consistency. I didn’t want to continue following anyone who hadn’t tweeted in a week. So, as I was going through my list, if a user hadn’t tweeted since the Super Bowl, they were gone. I also rid my following list of people who tweeted too much (yes, Guy, this is you). You can call this the Goldilocks approach if you want. Their activity had to be not too hot, not too cold, but just right.


After asking “do they use twitter?”, I then asked myself “do I care?”. This question is one of content. Once I narrowed my list down to people who tweeted frequently enough (or rarely enough) for my tastes, I then examined those who were left a little bit closer. Were they tweeting stuff (on average) that I cared about? Or was it all tweets about their personal family life, which, unless I personally knew their family, I had no business in anyway? This took a bit more time, as I had to go through user profiles and look at their general content. I also judged my connection with this person online, and gave a lot of weight to users that I’d interacted with online. If a user had retweeted one of my updates, or we had an online conversation, then that had a lot more weight.


After asking “do they use twitter?” and “do I care”, I had about 50 users I needed to unfollow. The last criteria I used was  ”do I know them offline?” This question is one of contact. My list contained some people that I had interacted with, and whose content was OK, but people I had never met in my life. In some instances, the first two criteria overrode this fact, but in many cases (at least 50) it did not.

Results (Less than 12 Hours)

So far, I’m LOVING the results. I can read every one of the tweets in my stream, without fear of missing something important. My interaction with the people I’m following has already increased. And that’s what this is about: people first (#ppl1st). Sometimes, that means having to get rid of some of the noise to maintain a strong signal. I’ll let you know further results as things develop.

Reblog this post [with Zemanta]

Leave a Comment