7 replies to this topic

[EGOL]

In the past I was running my logs through clicktracks to see how many visitors I get from twitter. To do that I filtered all traffic with "twitter" contained in the URL.

I was happy with the results... but today I thought... lots of twitter traffic goes through bit.ly and tinyurl. So, I reran clicktracks filtering for URLs containing "bit.ly" and "tinyurl". Found Nothing.

So, it looks like clicks from twitter to your site through one of these URL shortening services yield "twitter" as the referrer.

Anybody agree with that?

[eKstreme]

Correct. A URL shortener should not change the referrer if it does it properly (301 and, IIRC, 302 too).

BUT, when looking at Twitter traffic, there are two things to note:

1. Lots of bots keep an eye on Twitter in real time and request the links. So if you post a link to your site and it doesn't get shortened, then you'll see lots of those bots directly in your logs. On Cligs I try to differentiate between bots and human clicks, and it's a race keeping the bot detector accurate. These requests will have no referrer or a bogus referrer (like the URL of the app the bot is associated with).

2. Lots of people access Twitter using desktop clients like Tweetdeck and Twhirl. When these users click a link, it opens a new request in the default browser, and this will not have a referrer either.

The point is simple: if you see X hits from twitter.com, the real traffic because of the tweet is much larger than X.

Pierre

[EGOL]

Thank you, Pierre for the expert and detailed reply!

I didn't know about redirects carrying the original URL from twitter... and didn't think about the people who visit from desktop.

I appreciate your great response. Thanks!

[iamlost]

Pierre is quite correct: a significant percentage of Twitter users utilise third party apps showing no referer - just like direct traffic. Similarly desktop feed readers and iPhone apps to name just two. Direct traffic is a fast growing analytics black hole.

It becomes a matter of chipping away at the problem. A couple of examples:
* retain feed control - publish your own URL as the ONLY source feed. Do not send your subscribers directly to Feedburner

* tag your feeds, images, pdfs, etc. If you are really geeky these can be dynamically altered for every download. Or simply item specific. Strictly Business Capturing the Value of Content Marketing.

It is a matter of acquiring what data one can to differentiate user blocks. If one is unable to tell type-in from desktop program or 3rd party app in the middle how about learning what drew them back? That can be valuable actionable information.

In April bit.ly launched a voluntary app registration program: Registered Applications And Better Click Referrer Data, to add some referer value to their URL shortening service. An interesting idea that might have legs, only time will tell.

It would be far better if all these pigs in the middle were simply designed with referer passing in mind but as that is not about to happen anytime soon, if ever, one must make do.

[EGOL]

Thanks for the links, iamlost. I didn't realize that a person could redirect to feedburner. I guess I need to learn exactly how these things work. I think that feedburner is a great service and since it is owned by Google there could be more value in a growing list of subscribers than just the subscribers themselves - but one never knows?

[DrPete]

You could always do what the PPC crowd does and tag your links, to be sure. Add something like "?src=twitter" at the end, and it should come through no matter what (as long as the shorteners are parameter friendly). Shouldn't be an SEO issue, since you're only distributing these via social media.

Being a control freak, I decided to build a shortener just for my own site (sorry, Pierre ), and have started measuring those links differently.

[eKstreme]

Add something like "?src=twitter"

Two Cligs users created tools to integrate the short URLs with Google Analytics. The way it works is using the utm_* query string variables that GA understands. This is by no means unique to Cligs, but it's an easy way to do it

Tools:
1. Google Analytics Campaign URL builder with Short URL and Twitter posting.
2. The analytics issue with Twitter.

Shouldn't be an SEO issue

Long live rel=canonical!

I decided to build a shortener just for my own site (sorry, Pierre )

That's OK, Dr Pete. Next time I see you in a conference, I'll just accidentally spill tea on you

Pierre

[EGOL]

Great comments! And, thank you for the links to the articles.

Well, the only traffic you'll see coming from Twitter is traffic that comes through Twitters web interface. And most people I know don't use the web interface, they use desktop and / or phone clients. Because these clients aren't browsers, they won't pass along a referrer, and thus register as "direct traffic". (from Joost de Valk's article http://yoast.com/twitter-analytics/

Now I am starting to understand this.

Any ideas on how to distinguish bots from humans?



Pierre, Cligs is a great idea. Congratulations on making a nice product.