Google Analytics vs. Ad Server Reporting

All discussions in
Published by: Matt Pferschy , The Seattle Times Company
Published on: August 7, 2012

I know this is somewhat of an old subject and there are numerous blog posts and basic info available out there, but I am interested in getting some feedback from the field.

Google Analytics vs. Ad Servers

Over the past year and a half, we have experienced a noticeable increase in advertisers that are using Google Analytics to verify/validate our publisher side ad server metrics, particularly click-thrus. We have seen GA report as little as 1/3rd of clicks as visits. I have done a lot of research on why this may be, including the different counting methodologies (clicks vs. visits) and tracking mechanisms (cookies vs. IP and user agent)but it still seems almost impossible to me that there would be such a difference in the reporting.

For instance, if a campaign reports having 100 clicks in our ad server, but GA reports only 25 visits, that would indicate that each user clicked on an ad 4 times within a 30 minute time frame. That seems unlikely, and I can't imagine that the other issues that have been uncovered would result in such differences.

We have had some advertisers state that they are not seeing as large a discrepancy with other Publishers. So, I am working to discover if there are any patterns based on ad server, ad placement, rich media, etc. that could help uncover a source(s)of the problem and at least narrow the gap.

If any of you have experienced the same issue, I would appreciate any info you may have, along with what publisher you work with and what ad server you use. We use RealMedia 24/7's OpenAdStream Central as our ad server, and I know that a few other publishers that use OAS have experienced this issue.

Thanks in advance,

Matt Pferschy
Manager, Ad Ops and Analytics
The Seattle Times.

Comments

#1 Jakob Bak - August 13, 2012

At Adform, an ad-serving platform with a tag management platform that contains a site-tracking tool similar to Google analytics, we have been involved in at least 10 discrepancy investigations over the last years just like what you describe. While there are a lot of different and several persistent sources for discrepancies, some of the largest have often been related to un-intended clicks combined with the loading time of the advertiser website. Obviously rich media creatives with close buttons or placements close to menu bars etc. can tend to generate a lot of unintended clicks. Such is however only visible when deep diving into site-tracking data combined with ad-serving data, but this is rarely done by advertisers or agencies. But in cases when the advertiser site is loading slowly, or in particular that the Google analytic tag on the page is loaded slow, it becomes visible because many of the unintended clickers will have closed new window before Google Analytics has counted the visit, resulting in that click only being counted by the ad-server.
On the more positive side, we also see this same effect if the landing page is well known and clickers quickly click through to a second page. As an example, if the advertiser landing page is Facebook then a lot of people will be quickly clicking the “login” link before the Facebook landing page is fully loaded. So obviously if the Google analytics tag is not yet loaded, then all those clicks will only be recorded by the ad-server redirect.
I hope this might be helpful to you and hopefully unintended click is not the cause of your discrepancies. If you wish to test for this then I would suggest that you try to trigger a Google analytics tag together with clicks on one of the affected campaigns if this is technically possible, otherwise you would need to some set-up or tool that can measure on the ad redirects and site-tracking.
Br,
Jakob Bak
CTO, Adform

#2 Jalal Nasir - August 8, 2012

Matt

Since you are trying to verify clicks ( and not impression) I suggest using a service like bit.ly if you have the landing page url available or pixalate's click tracker for 3rd party ad tags ( if you dont have access to the landing page url)..

That might help you verify the clicks..

#3 Matt Pferschy - August 8, 2012

Thanks for your replies.

Jalal,
The metric in GA that I am referring to is just "visits" and not "unique visits".

Nate,
Our next step is to work with some sort of 3rd party in partnership with our clients, either using verification or a 3rd party ad server. At least that would help us tell our story and lend credibility to our reports.

Tom,
I would agree with you that GA is good for understanding trends and looking at the bigger picture. Our struggle is that there are many small to medium sized businesses who believe that GA is the authority and they often do not have the technical expertise or understanding to properly setup their UTM codes, etc, etc. We don't have this issue with larger companies that use agencies and their own ad servers. Our approach so far has been to educate the client as much as possible but we continue to get push back and have clients that stop advertising. That is very frustrating.

#4 Tom Good - August 8, 2012

Hi Matt,

We went through some big exercises with GA due to a few cases of variance with ad servers used. After a bunch of tests and use cases, there wasn't one factor that pinpointed the discrep. I'm sure this is in some other posts as well but we've had success with asking upfront if they are using any additional analytics tools to measure their campaign. If they are using Google Analytics specifically, we require that UTM tags are appended to their destination URL so we can at least parse our traffic out in their reporting. Not perfect realizing there are some inconsistencies, but it closed the gap to a manageable difference. Aside from that, it's a similar story to Nate above that GA is great for further insight into performance, but no substitute for ad server delivery or third party validation.

Tom

#5 Jalal Nasir - August 7, 2012

Matt

Question on this:

>>if a campaign reports having 100 clicks in our ad server, but GA reports only 25 visits

Do you mean unique visits?

#6 Nathan McCormick - August 7, 2012

Hi Matt,

Curious if you have tried other 3rd party verification services, ie AdSafe, DV, etc? Alternatively, have you been able to build credibility on performance results by using ad effective studies at all?

I ask because, while I currently use Ad Tech as my ad server I have used DFP in the past and experienced large discrepancies even between those two systems... Having used Ad Tech, 24/7, Zedo, DFA, DFP and Atlas, I have always seen discreps with GA and with no apparent connection to creative variance. We generally made the argument that comparing GA data is only a metric to further insight into performance. However, it is not a substitute for ad verification or delivery verification, etc that ad servers are qualified/certified to count. As such, they should not compare GA to ad server delivery. Of course an open dialogue is important to get to the root of what they have in question and perhaps other more effective 3rd party solutions can provide the "validation" they need.

Nate
Viggle