Comments on: Traffic Source Bounce Rates

By: Michelle

Michelle — Fri, 11 Jun 2010 12:42:40 +0000

Hi adam,
Wonderful post.,
One question regarding traffic sources..the referrer variable is supposed to capture the url of the page from where the user came from.But the referrer is counted even for page refreshes and back button is pressed..How to overcome this?
Would be really great if you can throw some light on this,
Thanks,
Michelle

By: Beth

Beth — Fri, 07 May 2010 17:19:43 +0000

Hi, can you still do the same approach at a keyword level?

Beth

By: Dan Cross

Dan Cross — Mon, 03 May 2010 15:11:14 +0000

Thanks Adam. I did check with customer care, and they confirmed what you’ve said here. Appreciate your attention.

By: Adam Greco

Adam Greco — Fri, 30 Apr 2010 17:36:45 +0000

I believe that your approach of Get&Persist would not work for Bounce Rate. I would use 2 sProps, one that does what you state and the second just for Bounce Rate. Feel free to check with @OmnitureCare to be sure though…

By: Dan Cross

Dan Cross — Fri, 30 Apr 2010 16:00:29 +0000

Hey Adam,
Great post, and very helpful. One question about this statement:
“In order for SiteCatalyst to know if a visitor made it beyond their first page of the visit (hence, did not “Bounce”), it needs to see a different value in the sProp at some point during the Visit. ”

Am I reading that correctly? Does this mean on a two-page visit in which a pathing-enabled prop was populated with the same value on both pages, Omniture will count that visit as “single access”?

We are planning to implement the traffic-driver-type prop you’ve described, but were planning on using “get&persist” to populate the prop on each page of a user visit. This would give us overall traffic numbers by source, but we also assumed would allow us to calculate bounce rates by source type too. No dice on that approach?

By: Adam Greco

Adam Greco — Fri, 12 Feb 2010 03:34:53 +0000

Maia – That is correct, but keep in mind that even if you use different identifiers, they can still go to the same variable (eVar or s.campaigns) unless you had a reason to put each into different variables (not recommended)! Using the getquerystring plug-in you can have as many identifiers as you’d like. It is totally up to you (and your developer!). That is why in the Paid Search Detection area, Omniture asks you to specify what your Paid Search string identifier is…While most clients use the same identifier for all campaigns, I like to use different ones for each. If you capture your fill URL in an sProp (which I also suggest), this gives you a great backup for segmentation purposes as well (i.e. URL contains “?=xid”). Let me know if you have any other questions…

By: Maia

Maia — Fri, 12 Feb 2010 01:33:26 +0000

Hi Adam,

I got here via your Tweet reply, in a different article of yours solving for another problem I have but I got struck by this post more. Will dwell on the other article later

I’ve been using only one call for external campaigns in my s_code and from what I’m seeing above, I can use multiple calls, is this correct? I’ve been confined to just extcid for all external campaigns and we just do classifications in Site Catalyst. You mean I can actually have different ***cid’s for my email, affiliate, etc. channels? How cool is that! I can then automate my classifications based on the unique identifier and not on the tracking code (which grows over time as more campaigns are added) for each channel which opens up a lot of free time for me

Am I reading this correctly or am I just overly excited?

Thanks,
Maia

By: Adam Greco

Adam Greco — Tue, 26 Jan 2010 15:38:04 +0000

I would not recommend trying to classify campaign + pagename which is why this solution is so much easier…

By: Adam Greco

Adam Greco — Tue, 26 Jan 2010 15:36:50 +0000

To calculate a site-wide bounce rate, I use the ExcelClient and divide Single Access by Visits.

By: Adam Greco

Adam Greco — Tue, 26 Jan 2010 15:35:19 +0000

Brad,

This is why you need to use an sProp for this. If you follow the instructions in this post you should be ok…