homepage Welcome to WebmasterWorld Guest from 54.211.157.103
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Become a Pro Member

Home / Forums Index / WebmasterWorld / Website Analytics - Tracking and Logging
Forum Library, Charter, Moderators: Receptional & mademetop

Website Analytics - Tracking and Logging Forum

    
Yahoo PPC - Alarming difference analytics & PPC reporting
Nashie




msg:3805240
 4:18 pm on Dec 11, 2008 (gmt 0)

Hi,

I have noticed a significant difference between the number of clicks that Yahoo is reporting vs the statistics Clicktracks and Google Analytics are giving me.

Over the same period:

Yahoo is reporting around 1800 PPC clicks.

Clicktracks is reporting a Yahoo total of 750, however when I segment that to split natural and paid search, that figure drops to give me around 355 paid Yahoo referrals.

Google Analytics gives me 739 Yahoo visitors in total (thats PPC & Natural).

As both of my stats above use JavaScript tracking, I thought I'd get the log files analysed, so another version of Clicktracks reported 501 Yahoo PPC visitors.

I realise that statistics will never be 100% accurate but Clicktracks Hosted, Clicktracks Pro, Google Analytics all suggest that Yahoo's report is wildly inaccurate.

Has anyone else experienced issues like this, and if so - any advice?

 

cgrantski




msg:3805557
 9:52 pm on Dec 11, 2008 (gmt 0)

To help answer, it's important to know exactly, and I do mean exactly, how you are tracking your Yahoo PPC and other visits. What is the raw material that Clicktracks or GA have to work with? How are these visits or clicks being marked as PPC or as Yahoo?

And, are CT and GA talking visits while Yahoo is displaying clicks?

Nashie




msg:3805965
 9:50 am on Dec 12, 2008 (gmt 0)

Hi cgrantski,

My Clicktracks data is collected using JavaScript tracking. All of my PPC ads user parameterised URLs in this format:

www.some-website.com/?r=yahoo
www.some-website.com/?r=google
www.some-website.com/?r=msn

In Clicktracks - I use the 'search report' which lists the unsegmented total number of visitors from Yahoo (750 visitors), and the number of Yahoo PPC (355 visitors) using my label which looks for the number of Yahoo visitors which entered the site with r=yahoo in the URL. I used the same method against my logfiles in another version of Clicktracks as well.

What is the difference between 'clicks' and 'visitors'?

Nashie




msg:3805981
 10:23 am on Dec 12, 2008 (gmt 0)

Ok, I just read the google adwords definition of differences between clicks and visits and I'm satisified that the scale of the difference between figures - (355 vs 1800) suggests that this issue can't be attributed to that alone.

cgrantski




msg:3806073
 12:34 pm on Dec 12, 2008 (gmt 0)

Thanks for the additional clarification and especially thanks for actually going to the documentation to answer that one question - a lot of people just lay their question out there and don't want to actually do any digging themselves!

At this point, if I were you, I would try to see if all the numbers are at least based on the same number of clicks. Forget visits and visitors right now. Can you get, from GA and/or Clicktracks, the numbers of hits (or "page views" or "views") for anything with "r=yahoo" in the URL? It doesn't matter whether "r=yahoo" occurs at the beginning or the middle of the visit - a full, simple count is what you need.

If that comparison of total hits is off by a lot, then there is indeed an underlying problem somewhere. One possible source of this problem is that not all of your Yahoo PPC clicks are coming through with that r=yahoo attached. This is something to check - both in your Yahoo setup AND in reality by clicking on a few of your ads. Another is that log file analysis may not tabulate two identical, successive hits as 2 separate counts. I'm not sure Google Analytics does either - it may count the second as a refresh of the first and leave it out. Of course, you want Yahoo to count a doubled click as 1, not 2, but I don't know if it does it that way.

If those numbers do match closely enough, then your problem is how the two analytics programs are dividing up those hits into visits, or into visitors. Does either program ONLY count a visit as being PPC referred if the referral marker (r=) is in the very first hit of the visit? That could be a big issue right there - all the visits where the PPC hit happens in mid-visit.

How are these programs sessionizing hits into visits? If with a cookie, good. If IPaddress or IP/UserAgent, then your "visits" number is highly suspect and probably much too low. GA uses cookies, but I don't know about ClickTracks.

Are any of your PPC ads going to domain names or landing pages that then redirect immediately to another page or domain? If so, it's possible that the r=yahoo is being stripped off.

Are you sure you are comparing identical time periods? If you're looking at a day of data, the different programs may be using different 24-hour periods ... using midnight based on Greenwich/Universal time, or the server's time zone, or your own time zone.

Nashie




msg:3806158
 3:00 pm on Dec 12, 2008 (gmt 0)

Thanks for your suggestions, and I am going to try those out shortly.

With regard to how Clicktracks is tracking, offhand I don't know but I can check - my assumption is that it uses cookies (is there any other way if it's hosted tracking?) as the combined Yahoo Natural and PPC figure of 750, is very close to the combined Google Analytics figure of 739 - so even without the PPC/Natural segmentation, the figure is still wildly out of kilter with the Yahoo PPC report.

The logfile version of Clicktracks revealed combined PPC/Natural Yahoo referrals of around 1000 which is expected when you consider that a certain percentage of users will have JavaScript turned off - again way way off the Yahoo report.

As a side note, the respective figures for Google and MSN all tie together quite nicely, there is no significant disparity with those PPC reports.

Interesting stuff, once again thankyou for your suggestions, it's always good to pick the brains of other clever people on this forum.

Nashie




msg:3873128
 11:29 am on Mar 18, 2009 (gmt 0)

After a significant period of time on the back burner, I've been looking into this issue again and I've found something interesting.

First of Yahoo acknowledged there was some clickfraud going on and credited the account, but what do you think of this issue...

As cgrantski was advising earlier I've tried to find statistics that match up more with numbers stated by Yahoo and on the Clicktracks site overview report when I look at 'Number of Visitors' it gives me the following for the week 8-14 March:

r=Google 13792
r=MSN 307
r=yahoo 2087

This is largely in line with the reports from the respective ppc admin tools:

Google = 14,719
MSN = 341
Yahoo = 2,271

But then I delved into the clicktracks search report:

r=google 11384
r=MSN 291
r=Yahoo 525

This is a puzzler. All of my ads are parameterized and the results on the 'site overview' seem to back that up as they are largely in line with expectation.

Clicktracks is recognising that around 2000 visitors come to the site with the parameter 'r=yahoo'... but is then saying that only 525 are actually being referred by yahoo.

Just to compare this against what Google Analytics says:

(Unsegmented unfortunately so this includes both natural & PPC)

Google = 14,363
MSN = 702
Yahoo = 848

Now based on my rough percentages of natural traffic, around 35% of the yahoo referrals were natural which keeps things roughly in line with Clicktracks search report.

The question I am now asking myself is - how are Google Analytics & Clicktracks working out the identity of the referring server, it almost looks as if both Google Analytics and clicktracks are not capturing this information (although where the blame lies for this I would not know).

I hope I've made some sense during this post, the volume of stats I am throwing about can be a bit overwhelming at times. :)

cgrantski




msg:3873179
 12:20 pm on Mar 18, 2009 (gmt 0)

Hi Nashie --- First of all, Yahoo etc have networks of search engines so any number of your clicks, administered through the Yahoo PPC program, can come from other places.

Second, referrers are often absent, for a big variety of reasons, even when there WAS what you'd call a referring site. Do a search on something like "direct traffic no referrer" for lots of discussions about all the reasons. There's a WebTrends Outsider blog that has a pretty complete list that I refer to, and probably plenty of others just as complete. It's been discussed here in WebmasterWorld also.

Trust your parameters. If you have the right reports, they'll give you an idea of the real source of all those syndicated ads (if they can be identified at all), and that can be eye-opening.

Nashie




msg:3873401
 4:15 pm on Mar 18, 2009 (gmt 0)

Interesting. I was thinking about how that would effect my natural search tracking. I'll just still use the search report for that, the absolute numbers may be inaccurate but the proportions involved should still be valid.

Cheers cgrantski.

Global Options:
 top home search open messages active posts  
 

Home / Forums Index / WebmasterWorld / Website Analytics - Tracking and Logging
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved