I’ve been thinking about how one could measure noise on Twitter. By noise, I mean people twittering without building an audience. If a person talks and talks and talks but no one listens, that’s just noise. At least, that’s my working definition of noise on Twitter.
Working off that definition, I decided to take a look at the MN tag on Twitterholic. That tag on that site lists people who’ve defined their location at MN, so a subsection of Minnesota based Twitterers. This doesn’t include people who use Minneapolis, MN, other cities, city/state combinations, geocodes, or Minnesotans who don’t disclose a location, so it’s really only a small subset of Minnesota twitterers.
If we look at followers, the top-5 most followed accounts right now are:
1. Steve Gewecke (mlmphonepro)
2. Mark Salinas (Mark_Salinas)
3. swamibaby (swamibaby)
4. Nicole (niclone)
5. Michelle Judd (mjudd)
That says something about the quality of their tweets, although there are ways to game follow counts, such as running follow-bots that automatically follow other Twitter users under the assumption that a portion of them will follow you back.
Turning to noise, my working definition for Twitter noise, is the Update to Follower ratio. By taking the number of times someone has tweeted and dividing that into how many followers a Twitter user name, you can get a sense for whether the user is building an audience based on what they’re tweeting. So if someone has twittered a ton but has relatively few followers, they’re noisy by this definition. On the other extreme are people who have huge following yet have tweeted few times. They’re often celebrities like Oprah.
Here is what noise looks like among the top Twitterholic ranked Twitterers who’ve labeled their location as MN:
That, to me, is a lot of noise. Five Twitter users combined have published more than 479,000 tweets. What’s really impressive is that these accounts are only 6 months old. That is a lot of twittering. After twittering 100,000 times, you’d think a person could build a following larger than a few hundred.
A local Twitter measurement service called FollowCost helps determine what you can expect if you follow a particular Twitter account. Here’s what they say about the #1 noisiest Twitter account.
For those of you not familiar with Scoble units, they’re named after a relatively heavy Twitter user, Robert Scoble. As the above graphic illustrates, @HouserwarePlaza is historically 31X noisier than than Scoble, although their last 100 Tweets have been a mind-blowing 42X noisier than Scoble.
By the way, the stats for these 5 Twitter accounts are already way out of date. I grabbed those numbers on December 27th, and the accounts have spewed out around 3,000 tweets each since then.
What’s going on here? Is someone really Twittering 1000’s of times per on these accounts? No. Instead, an online retailer in Brooklyn Park is auto-tweeting their website’s inventory over and over again in an attempt to show up in Twitter search results whenever someone happens to be looking for a “Range Kleen P104 P-1048″PORCELAIN STICK-FREE PAN” or “OMCAN BSB322104 104 BAND SAW BLADE 322“.
It’s going to be interesting to see how Twitter addresses accounts like this in the future. The service is under no obligation to allow thousands of robotweets into their network per day. It costs money to store all of that inventory. And, more importantly, Twitter could determine that there isn’t enough value to their network to support this sort of thing.
If that happens, Twitter could either block account like this (I believe that does happen from time to time) or perhaps they could charge a nominal fee per daily tweet over a certain threshold to help recoup their costs or slow down extraordinarily heavy users. At a penny per tweet, would it still make sense to use Twitter in this manner?