Omniture SiteCatalyst Implementation Question One- Capturing Impressions for Different Ad Types

Naysayer: Two blogs in week?  Couldn’t be at tooltime.typepad.com.  All that guy does is surface every couple of months to bash Omniture and put me to sleep with the nuances of report building. 

Jesse:  In spite of the naysayers out there, here is my second blog in less than a week.  Some might say that I am doing this only because I have content to repurpose (true). Others will claim it is only to boost my traffic so I can play with the cool new features of Google Analytics (truer).  A few might speculate that it is to hock a few more Semphonic SiteCatalyst Implementation Toolkits (truthiness).  But the real reason I am doing this is to selflessly give back to the web analytics community (false).

Prior to this year’s X Change conference Semphonic announced a contest where folks in the web analytics community could submit questions about the toughest part of an Omniture implementation.  The prize for the top questions would be a copy of the Semphonic SiteCatalyst Implementation Toolkit and all questions, winner or not, would receive a detailed answer from a Semphonic consultaint.  In my next three blogs I will post the three winning questions as well as the Semphonic answer to each of them.  The winning questions were selected for both the implementation challenges they posed and because of their relevance for a wide audience of SiteCatalyst users.  Hopefully someone out there is facing a similar challenge and finds one of these posts relevant and helpful. 

Without further ado here is the Q & A for one of the top three questions.  You can expect the other two to be posted over the next few days.

Question:
How can I capture ad impressions for six different types of ads, such as paid hero, internal tower, etc., into six different Omniture variables?

Answer:
You most likely can populate a prop and/or evar variable that fires on page load when a certain ad type is detected.  Implementing prop and/or evar variables would be easiest and cleanest if you have existing page coding that can identify ad types.  In the case that this is not available, you could try hard coding the variables.  This approach works if each page on your site is set up to consistently populate ads in the same location(s) upon each load.  In a site where the same page will sometimes load with an ad and sometimes without, you will have to implement a check to determine if the ad exists prior to sending (or not sending) the hard coded ad tracking.  Unfortunately, if there is no consistency of ads populating on a page and no way to determine within the coding if an ad has populated or not you’re out of luck and won’t be able to get the impression tracking you desire.  

Assuming it is possible to implement tracking, the impression variable(s) could be placed in an assortment of Omniture reports.  You could populate a prop and/or evar report with the ad types and the impressions associated with them.  This prop and/or evar report could look like:
A. Ad type (i.e. tower)
B. Ad type and paid/house (i.e. tower-paid)
C. All ad types and paid/house on page (i.e. tower-paid_hero-internal)

You most likely will want to implement tracking at the most detailed level you would like exposure to.  Given that you are only interested in impressions, implementing detailed tracking and using SAINT to roll up to more general categories, like ad type, will give you accurate granular and general impression tracking.  If deduplication of visits or visitors might be a future need you should consider implementing impression variables with varying levels of detail in different prop and/or evar slots.  Any prop(s) and/or evar(s) should have relevant correlations/subrelations enabled to enhance the analytic capabilities of these reports.

As an alternative strategy, it would be appropriate to utilize an evar counter for each of the ad types.  This could give breakdowns of the quantity of impressions by visit, visitor or by custom metric like paid search keyword.

Comments

[Adam Greco]

The following is an alternative approach. I would pass a list of ID's corresponding to the six ads presented on a page to the Omniture Products variable (separated by commas) and set an "Impressions" success event. That would allow you to pull a report that shows how many impressions each ad ID received. Then when a user clicks on one of the Ads, I would capture that Ad ID in the Products variable (by itself) and set a "Clicks" success event. This will aloww you to create a calculated metric that divides Clicks by Impressions to see the Click-through rate for each Ad ID by opening the Products report using the two success events just mentioned. At the same time, I would pass the ID of the Ad that gets clicked to an eVar so that you can see what website success events took place after the Ad was clicked (since the Products variable isn't persistent). Finally, as was mentioned by Jesse, use SAINT to classify the Ad ID's in both the Products varioable and the eVar to group Ads into buckets (i.e. Hero, Tower, etc...). This approach should limit the number of variables needed and takes advantage of the full subrelations of the products variable, while providing the desire data...