The analytics issue with Twitter

Great idea! Twitter traffic can (and should) be measured properly with the right approach.

Do these parameter additions not mess somehow with the tweetbacks code? I'm already having trouble with it - don't want to court more trouble!

Also, how do you get your RSS tracking into the RSS feed? Both seems to be very useful techniques.

For an interesting look at follower counts I set up a website that I planned to promote solely through Twitter, http://staires.org, a daily music thing. Admittedly there are a few things popular twitter users do that I don't: I'm not big on conversation (though I respond and try my best I don't strike up convos) and my content is very much "you can look once and determine if you care to click"... But I'm up to 630 followers now and I get about 20-30 actual visits a day. I have very little (almost 0) search engine traffic even after unblocking them in my robots.txt. Since this a post about analytics and if Twitter generates real traffic I figured I'd share my "high follower count" story.

Followers != Clicks.

The websites I manage use OneStat, is there a way to track Twitter usage on that platform?

Joost, thanks for a great solution to twitter referrer stats!

As implemented, a visitor may end up bookmarking (delicious, digg, etc) the URL with the campaign stuff in it. This would mess up your stats. To avoid this, include this script below the Google Analytics script. It modifies the browser history. The reason the RegExp is complicated is becase, as Joe points out above, there may be other parameters in the URL that you want to preserve. Disclaimer: I haven't widely tested this but it does work in FF2 (linux), FF3 (win32), IE7 with a handful of URL variations. Also, my javascript is weak so there may be a better way to implement this.


// Show a clean URL if this came from a tweet/cli.gs URL.
var campaignRe = new RegExp(/([?&])utm_campaign=twitter&utm_medium=twitter&utm_source=twitter(&?)/);
var loc = new String(window.location);
var match = campaignRe.exec(loc);
if(match != null) {
// Remove the crufty URL from browser history.
if(match[1] == "?" && match[2] == "&") {
// Change & to ?.
var cleanLoc = loc.replace(campaignRe, "?");
} else if(match[1] == "&" && match[2] == "&") {
// Preserve &.
var cleanLoc = loc.replace(campaignRe, "&");
} else {
// Just drop the params.
var cleanLoc = loc.replace(campaignRe, "");
}
window.location.replace(cleanLoc);
}


@Gif Dump: does this solution change your opinion about "bad SEO written all over it", or is there still some problem I'm not seeing? Can you elaborate about what you mean? The crufty URL is cloaked by cli.gs so I don't see the issue.

just came across this site, thanks joost. great stuff and this will be a great addition to our twitter unknowns!

I would say you would be better off specifying:
utm_medium = referral
utm_source = twitter.com
utm_campaign = Tweets

That way you can lump all the referrals together appropriately, lump all the twitter.com referrals together and just analyse it based on "Tweets"

FWIW: I wrote a plugin for Mint that helps you track just these kinds of campaigns in realtime: http://craigmod.com/interactive/link_spice/

Nice bit of info. :-)

I thought there would be some tracking for those visitors who are visiting my twitter profile

Great information as alway, I am going to try this on my site because it's almost impossible to track twitter traffic as you have so correctly pointed out.
Thanks again.