The media loves to hate Twitter when it comes to measuring their “influence”, or how many people click links shared on Twitter. Even when they state the obvious reasons (HTTPS sends no referrer, 99% of people use Twitter “apps” which also send no referrer, etc), they brush them off and pretend no one uses Twitter.
I’m not here to defend Twitter, but they just made a big change that almost 100% guarantees headlines like the one we’re using here.
What am I blathering on about? I can’t exactly tell what Twitter changed, as I never paid attention to t.co really, but either they finally started actually routing every link through their shortener, t.co, or they changed the way t.co redirects. The bottom line is this: you will pretty much never see a single referrer from twitter.com again. Instead you will be seeing t.co. (At least from non-Clicky analytics services – more on that in minute).
The reason is because of the way t.co redirects. Instead of using a 301 redirect, which every shortener should do, they are using a combination of a “META” redirect and Javascript to send you on your merry way. If you click a t.co link and hit escape really fast, your browser will pause on that page, allowing you to see the source:
<noscript><META http-equiv=”refresh” content=”0;URL=http://www.addedbytes.com/blog/if-php-were-british/”></noscript><script>location.replace(“http://www.addedbytes.com/blog/if-php-were-british/”)</script>
The “problem” is that both META redirects and Javascript redirects overwrite the original referrer. The referrer becomes the page with that code on it. Hence, no more twitter.com referrals, because every single link passes through t.co now.
Here are screenshots of our 30 day history of referral traffic for twitter.com vs t.co. Notice the last 3 days, twitter.com is zero, and t.co has come out of essentially nowhere:
What to do?
We just updated Clicky to convert t.co referrers into “twitter.com” automatically. Knowing the specific t.co link someone came from has no value, but knowing someone came from Twitter does. No other analytics service has done this yet that we know of, so as usual our tiny size lets us move lightning fast to address issues like this.
[Update: ok, we changed it back. See comments.]
But wait there’s more! There’s a damn nice silver lining here. The main reason twitter.com doesn’t show up as a referral much, even before this change, is because so many people use apps instead of the actual twitter web site. When you click a link in an app, that app passes the URL to your browser, which then opens it – but that means no referrer.
However, now that all links are being passed through t.co, and this happens via the API as well (I confirmed via our own Twitter keyword monitoring feature), this means that even people who click links from apps will be passing through t.co first. If we automatically convert t.co links into twitter.com in the backend, this means we will be able to give you a MUCH more accurate picture of Twitter’s traffic to your web site. (We just pushed this change so we can’t guarantee this will be the case, but based on what we’ve analyzed, if you use Clicky you should start seeing a huge influx of twitter.com referrals).
This may even be the reason that they’re doing redirect this way, because they’re sick of the uninformed media saying they have no “influence”. By having a ton of t.co hits suddenly showing up in referrer logs, maybe the media will consider what that means. Doubtful, but one can always hope.
