Tag Archives: advice

Good example in the last 24 hours of why we always need to question statistics.  Since these concern Twitter, I’ll keep this brief although not 140 characters.  Multiple sources got themselves in a frenzy yesterday buy publishing, as did one source, the following:

Traffic to Twitter continues to decline as the microblogging site seems to have passed its peak in popularity, according to data aggregated by eMarketer. Nielsen has traffic to Twitter.com down by a dramatic 27.8% between September and October 2009, falling to 18.9 million unique visitors. comScore claims unique visitors to the site are down by 8.1% in October. Compete reports a 2.1% decline during the month.

Wow.  Twitter has jumped the shark.  Short the stock.  Number don’t lie, right?  Well, if you read the same source this morning, wrong:

Disclaimer to those Twitter stats we published yesterday: the figures just include unique visitors to the Twitter web site and do not include third-party application or mobile phone usage, according to Nielsen. That’s not to say that apps that feed into the Twitter community such as TweetDeck, TwitPic, Tumblr, etc. aren’t also losing followers, however. Nielsen studied audience retention on a number of those apps last April and found that about 60% of users who try these services end up abandoning them after one month – about the same abandonment ratio as on Twitter.com.

Well, we all know that there is a large abandon rate for many new apps and services as users try them until them settle on the one with which they’re most comfortable and which accomplishes whatever task to their satisfaction.  I’ve been through a bunch of Twitter interfaces just as I’ve tried and abandoned several niche social networks, different browser add-ins, and even makes of PC over the years.  One quick look at the tweets I get from the folks I follow shows a wide diversity of clients, very few of which are the Twitter interface itself.

The lesson?  Don’t take research at face value.  Ask questions, especially the ones not asked in the research itself.  Those are my thoughts.  What are yours?