I noticed in the past two or so months, after Google upgraded their SERPs to enable users to rearrange and delete results (see thumbnail to right) that the outbound links Google was posting on SERPs were redirects through the google.com site (meaning, they were tracking clicks on outbound links.)
Since I often search for something on Google, then “right-click”, “Copy” to get a link, I noted the links were not the actual links, but Google.com links. This went away shortly thereafter.
Today, checking again, they are using JavaScript to track outbound links (View Source on a SERP and examine the outbound links).
For example, one of our decent organic result page for the search “ROI Tracking” shows us as:
<a class="l" onmousedown="return clk(this.href,'','','res','4','')"
href="http://www.conversionruler.com/">Measure and manage your
PPC Marketing Campaign Performance with<b>...</b></a>
For this post, I’m going to focus on the technical breakdown of what “clk” is tracking.
clk generates an image-bug tracking call. The parameters to clk are:
- 1st parameter: The URL you will be sent to (this.href)
- 2nd parameter: Original Intention (Guess?) (”)
- 3rd parameter: Referring domain (from toolbar) or affiliate (Guess?) (”)
- 4th parameter: Where you clicked (res for Search Results, clnk for Cached links)
- 5th parameter: The position of the result you clicked on (1, 2, 3, 4, etc.)
- 6th parameter: “Extra” parameters to pass (probably for future compatibility or labs) (”)
Based on the JavaScript code and a (very) cursory review of the Google site and our analytics logs, here’s an attempt to decipher what they’re tracking.
What Google is tracking
Google tracks your link with a query string to:
http://<google domain>/url?sa=T&source=web&oi=(Parameter 2)&cad=(Parameter 3)&ct=(Where you clicked)&cd=(Search Position)&url=(URL to be sent to)&ei=(Security Code)
Here’s the breakdown:
| Google Parameter | Meaning | Examples |
|---|---|---|
| url | The url you are clicking on | http://www.conversionruler.com/ |
| sa | Big guess. The tracking function on the Google site to invoke. |
N,X,T
X appears for spelling corrections |
| source | Where the search originated from. | web |
| oi | Big guess. The original intention of the search? |
spell when the user misspells a word and corrects it with “Did you mean …” coopctx found in log files. revisions_inline when the user clicks “Searches Related to” on the SERP page |
| cad | Big guess. Appears to be a referrer, affiliate, or client token. |
So far only seen with google maps, it is a number (6,7,8, etc.) which I’ll assume are Google internal map application identifiers.
It appears to be the search origination, when it begins with a letter, it’s a site: cad=Tvtrip.com (extensive client of Google maps) I have a feeling that when Google Toolbar automatically extracts an address from a web page and forward it to maps.google.com, this is the original domain. |
| ct | The click location. | res for results clnk for Cached links in results other codes TBD |
| ei | Security code, encoded user identification, timestamp, or some combination of all three. | This value is identical for all results on a page, but appears to change page to page. |
| cd | Search result position | 1,2,3,4, etc. |
The parameters here apply for many of Google’s Referring URLs as well, so feel free to use these definitions as you see fit.
If you find this useful, please link or comment. If you have further information, I’d appreciate it if you’d share in via inbound links or comments.
11 replies on “Analysis of Google outbound link tracking”
nice article
A related article which is useful regarding referrer URLs:
http://www.blogstorm.co.uk/google-adds-ranking-data-to-referrer-string/
[…] of the various parameters being used and newly introduced, Kent Davidson had put up this post Analysis of Google outbound link tracking about google’s use of JavaScript to track outbound links from the search results. He has neatly […]
nice and will it change again ???
when will they have the live search, bing is surely the way forward.
with google tracking browsing behaviour, does it affect my page ranking? is there any way to manipulate google by faking behaviour?
Very good article, and Yves thoughts are the same as mine. Could this be a new way to manipulate/fake behavior?
Great article, Thanks for sharing.
[…] with outbound SERPs. More like outright usurpers if you ask me. Check this Razzed blog for further Google outbound link tracking analysis. And try to pay attention where you’re sending people with your copied links. I sure will […]
I noticed &cad=h when searching today. It showed up while I was in Chrome and happened after copying the search terms from one tab to put them into another tab to search.
I’ve noticed the “ei” code before. As you say it is unique to a page hyperlink. You would think that when you click the “Next” button, you would use the same “ei” code but it uses a different one. Would be interested in any more that you find out about this.