Update 6/5/2009: Added Bit.ly/Google comparison charts
Update 6/8/2009: Added caveats/clarification on data pulled
In pulling together some data on twitter clickthroughs, there was a noticeable discrepancy between bit.ly and google analytics data - bit.ly clicks are noticeably higher than traffic source numbers for twitter on Google Analytics. Two caveats:
1. Bit.ly data collected reflects all bit.ly links including those initially posted by @worldresources using the Google Analytics Campaign Code - accounting for an average of 60% of the clicks - as well as those generated by others.
2. I pulled all information by keyword "twitter" from the Google Analytics account - which brings over all pageviews sources coming from twitter and the GA Campaign Code links posted by @worldresources.
But in closer inspection, I noticed that the Bit.ly clicks and Google pageviews data lines - despite being vastly different in terms of numbers - were almost identical in terms of trends.
I did some research on this and here is a list of potential reasons for this -
- No two analytics tools measure data the same way - there will always be discrepancies between analytics tools. With bit.ly we are measuring clicks but with google analytics, we are measuring pageviews which may be captured differently.
- Google Analytics may be combining multiple clickthroughs by the same twitter/user account as one whereas bit.ly counts clicks and not users
- One prevalent theory is that bit.ly numbers are higher because bit.ly data includes bots and automated traffic whereas Google Analytics will not capture those visits (in GA count is computed using Javascript, which bots do not execute).
- In the March 2009 post by Om Malik (Why Bit.ly Will Upstage Digg), he explains
"...with bitly it doesn’t matter where those URLs are embedded -- Facebook, Twitter, blogs, email, instant messages, SMS messages, desktop clients -- a click is a click and Bit.ly counts it, in real time."
So while some of the referrers may be bots, because twitter feeds are disseminated throughout the web and desktop clients, not all of the clicks are going through twitter. Aha! Note that as of March 2009, bit.ly screened out HEAD requests from click results.
- Another theory is that Google Analytics does not count clicks from twitter.com as uniques -- and that's one reason google counts are lower than bit.ly. It isn't clear how traffic from twitter clients looks to GA, but this may account for overall lower count rates on GA.
- Not all bit.ly clicks are coming from WRI-generated links, and these charts only includes bit.ly data - it doesn't include click on links using full WRI URLs OR other URL shorteners like is.gd, tr.im, cli.gs, tinyurl.com, twurl.com. I haven't determined how to capture that additional data efficiently - I'm still trying to get my mind around the bit.ly/google data.
Regardless of which numbers you ultimately decide to highlight (Google Analytics Traffic or Bit.ly Clicks), you should track twitter links in Google by including the Google UTM campaign code as part of the link - see the EpikOne post Twitter and Google Analytics: What to Track by Justin Cutroni.
