Measuring the amount and quality of traffic generated through Twitter can be challenging since most Twitter activity occurs outside of the web, on desktop clients like TweetDeck, mobile apps like Echofon, and other types of third party applications.
I’ve found six reports for Twitter client statistics from Many Eyes/IBM, TwitStat, Sysomos, Twalytics, Funkatron, and TweetStats, however there are some big differences among their reports. For example, web (twitter.com) usage varies from 18.9% to 48%; TweetDeck varies from 6.6% to 13.8%; and Hootsuite varies from 0.6% to 5.2%. That’s a lot of variation considering these reports were all generated on the same day (Feb 5, 2010) with the exception of the Sysomos report (Nov 16, 2009). Averaging the results among all six reports reduces the variances likely due to the different data collection methodologies.
When we average the results we can see that 38.1% of all Twitter activity takes place on twitter.com (Web), and 2.4% takes place on m.twitter.com (mobile web), for a total of 40.5% of twitter activity coming from the twitter.com domain. Why does this matter? Marketers need to be able to measure the volume and quality of traffic coming from Twitter to determine an ROI for their efforts. If you’re only measuring visits from twitter.com in your web analytics tool, you’re only measuring 40% of all the activity.
Here are three possible solutions…
1. Trackable URL shortening services
Services like bit.ly, cli.gs, tr.im, and HootSuite’s ow.ly will provide statistics on how many clicks were generated from your twitter posts. This is especially useful if you just want the number of clicks, but any good analyst knows that it’s what happens after the click that really matters. These services cannot measure any of the activity that happens on your website post-click, so using URL shortening services alone can only provide limited insights.
2. Append tracking codes to all links posted on Twitter
Most web analytics tools, like Google Analytics, allow you to append tracking parameters to links in order to report certain visits under a specific campaign rather than direct traffic. Links posted to Twitter are perfect opportunities for tracking codes since around 60% of your Twitter traffic could potentially be reported as direct visits. If you use Google Analytics, Google offers a free tool to help you add these parameters to your links.
So instead of posting your link as:
http://www.mydomain.com/abc
You could use this link instead, with added Google Analytics tracking paramters:
http://www.mydomain.com/abc?utm_source=twitter-feed&utm_medium=social-media&utm_campaign=twitter-abc
These parameters can make links really long, so you’ll need to use a URL shortener to post them on Twitter. When you add tracking parameters to all links you tweet, you can start to measure the actionable fun stuff like goal conversions, time on site, visitor loyalty, visitor recency, and bounce rate for all traffic from Twitter, rather than just 40% of it. This information can be a starting point for generating an ROI on your social media efforts.
If you give utm_source a value different than “twitter.com”, you can be sure that all visits from “twitter.com” originated from links posted by tweeters other than you (a good thing!). Comparing your own efforts (traffic from the twitter-feed campaign) to your fans’ efforts (traffic from twitter.com) can yield some pretty neat insights. This can be a good way to gauge how energized the twittersphere is about your brand. To get the most accurate view possible, I’d recommend using a multiplier to estimate the total number of visits from twitter.com (details below)
3. Use a multiplier to compensate for visits from Twitter clients
Since 59.5% of all Twitter traffic could be reported in your web analytics tool under direct visits, a quick (and dirty) way to estimate the number of visits you’re really getting from Twitter is to multiply the number of twitter.com visits by 2.47 (which is 100%/40.5%). This calculation assumes that your audience’s twitter client usage matches that of the overall Twitter population (40.5% use the website), and that the audiences among different clients are uniform (TweetDeck’s users behave exactly like UberTwitter’s). Of course, this is likely not the case, but the more traffic you receive from Twitter, the closer this estimate will be.
Using tracking parameters is definitely the better way to go for accurate & insightful Twitter stats, but the multiplier is a good starting point. And as explained above, tracking parameters can’t be appended to links that are posted by your fans, so applying a multiplier is necessary to estimate the total number of visits generated by all tweeters other than you.
