How to track Universal Search traffic with Google Analytics

How to track Universal Search traffic with Google Analytics
Originally posted on January 16th 2009 on the Searchcowboys website but has moved here since May 2010.

After reading the post by Heini van Bergen on setting up filters on Google Analytics something caught my attention. Something I was looking for, since I started researching Local Search Optimisation.  I wanted to know how many people were actually clicking the blended results you see in Universal Search. Google Analytics measures these clicks as Google (organic). But what if my link appears in the SERPS twice? How do I know which link was clicked? And if my link is listed in the SERPS more than once, which one of them is converting in a better way?

Image: Universal Search illustrated.

Looking into the SERP source code can give you a lot of insight. Google is showing parameters for
Universal Search results. This will give us the possibility to set up some filters or advanced segments, based on the full referrer URL.

Overview of parameters:

  • sa=x eXtended listings (regular tracking links show sa=t , traditional )
  • oi= this parameter specifies the specific universal search feature.
  • cd= internal position of a link within the specific universal search item specified with `oi`.

Unfortunately not all links from Universal Search boxes can be monitored in this way directly, for example those listed from Local Search results.
These will show a direct link to the landing page designated for the local business owner. This landing page can be defined in the Local Business Center (and you should!).
So what can we measure directly and how do I set up filters or create advanced segments for them?

Measured directly

The following are specified by the parameter oi

  • Blog Search: blogsearch_group,  blog_result
  • Image Search: Image_result ,  image_result_group
  • Spelling corrections: spell
  • Sitelinks: smap
  • Definitions: glossary_definition
  • Suggestions (link to suggestions/revisions): revisions_inline, revisions_narrow

Measured indirectly
this means that a user clicked on a universal search link, but went to the Google landing page first before going directly to your website, parameters are passed on by Google. Note that if this occurs, it can mean that it took one click to many for the user to get to your website because they couldn’t find what they were looking for in the first place.

Setting up the filters

Step 1 :
Create filter for organic traffic
New filter ->
Filter name `Organic`
Filter Type -> Custom Filter
Checked : include
Filter Field: Campaign Medium
Filter Pattern `organic`

Step 2:
Create filter for All Universal Search traffic
New Filter: ‘Universal Search Items
Filter Type: Custom Filter
Checked: Advanced
Field A -> Extract A -> Referral -> (.*)oi=([a-zA-Z_]+)&(.*)
Field B -> Extract B -> Referral -> (\?|&)q=([^&]*)
Output To -> Constructor -> User Defined -> $B2 : $A2

You can also create separate filters for some of the Universal Search items instead of all of them.

Custom Step:
Create filter for Universal Search – specific
New Filter: ‘Universal Search | images’
Filter Type: Custom Filter
Checked: Advanced
Field A -> Extract A -> Referral -> (.*)oi=image(.*)
Field B -> Extract B -> Referral -> (\?|&)q=([^&]*)
Output To -> Constructor -> User Defined -> $B2 : $A2

You can repeat the custom step above with all the variables for ‘oi’ available.
The output (after step 2) should look like this:

Also, you might consider filtering out the &gl and &hl parameter so you know which engine the search originated from.

What can we do with all this information?

Nice to have all these filters setup and to gain some more insights in clicks from the organic search results, but what use has it?
If the search results for a specific key phrase show both images and ‘regular’ text links for a specific key phrase, how can I tell if the image or blog link was clicked instead of the regular one? They both seem to originate from the same source (Google), same medium (Organic) and for the same keyword. Setting up these filters will help you distinguish the traffic. (It will give you the 3-click-path in some occassions. Thx for Tom Critchlow for clarifying this. )
Do images get less clicks? Or more? Or are users more likely to click the blog result than the regular search result? And is someone who clicked on an image more likely to convert into a lead or action? Or is it the opposite?

Set up these filters and you will know! Feel free to share your experiences.

Leave a Reply