Partner pages - Page name inconsistency

[ClaudioMa2]

Hi everyone,

on the pages linked to the partner program  a concern has been raised about the inconsistency of the valorization of the variable v84 Partner Name.

The v84 should be valued with the name of the partner to which the page refers but the number of page views is different whether we consider the page name once or the v84. (8% discrepancy).

The aim would be to find out which of the two tables is reliable (or more reliable) and understand possible reasons of this discrepancy. Suggestions?

Thanks

[abhinavpuri]

Hi @ClaudioMa2 

There are not much details available but there may be scenarios where eVar84 would persist beyond the partner pages and would report page views under last value tracked/persisted.

The pageName is much more consistent when the goal is to track page views as it will not persist beyond hit tracked.

Another way is to use eVar84 instances metric with eVar84 as dimension. Assuming eVar84 is only tracked on pageview calls and not custom link tracking calls.

[ClaudioMa2]

Yes, I used it that way. The problem is that the page views are 8% more than the partner name instances(v84) with full page name as dimension or partner name(v84) as dimension. Thanks

[Damonwhall]

@ClaudioMa2 , I often audit dimensions that are supposed to be parallel to each other.  

In this case, I would create a venn view, drop in "Partner name = Bob" , then drop in "page = bob's page" and then use "page view" as the metric.  

If there you see anything besides a perfect overlap (meaning that the sizes might be different, but still within one another) , then you have removed one of 2 problems (where the partner eVar is firing incorrectly or persisting incorrectly).  

(note: if you do see part of the partner eVar fall outside of the venn overlap, then create a segment that is "partner name exists and does not = bob and page = bob" and then see in a table what pages show under that segment.  This will allow you to dial into the page culprit that is leading to a discrepancy.  )

Now that you have an overlap but a difference in volume, I would create a segment that is "page = bob and partner does not exist" and see what that volume is.  That will show you the count of how pervasive it is and allow you to see other dimensions like time on page to determine if there is something unusual about the related audience.  

Lastly, I would test the experience myself and see if the proper dimensions are firing when I'm on the site live.  capture your ECID (hopefully you are capturing ECID in a dimension) and use that to follow yourself in a flow chart to see if all of the proper steps are being captured.  

That will let you know if the issue is pervasive.  Then, I would work with your dev in Launch to ensure there is not a race condition or something regarding consent and data capture that is causing this.  Example, you might find this is only happening on hit 1 and not other hits (the EMEA might be the cause)

[Jennifer_Dungan]

This is good advice.

Also, is there in the implementation do you have mobile apps, or imported data sources that could be missing the eVar? Also, are there potentially any Processing Rules or Vista rules that could be removing the eVar under specific scenarios?

The Page Name (Page) dimension has a hit level attribution, and your eVar is either set to Hit or a longer Attribution.. so to be seeing higher values on your Page Name (unless of course you are using a custom eVar for this as well with a different attribution model to v84) seems to me that either your eVar is failing to set (either from the website tracking or potentially from another data source), or it's being removed through some sort of post processing?

If you can identify the pages where eVar84 is missing, you might want to try and correlate other dimensions to try and see if there is any sort of pattern or scenario that could be leading to the missing data.