Server logs : PPC visitors

[stevew]

Hi,

I use WebTrends software to analyse our server logs.

Until now, I've assumed that PPC visitors were included in the "Visits Referred by Search Engines" (VRBSE) figure.

This afternoon, I've gone back a while and found that about a year ago, the actual number of PPC visitors was greater than the VRBSE figure!

I've run some test visits, looked at the logs and the log file information seems to indicate that the right information is being recorded for PPC visitors.

I use tracking urls (eg adding "?src=g&c=ms" to the target url) for all PPC campaigns, but I wouldn't expect this to prevent it registering as a Search Engine referral.

Has anyone any ideas?

[pielover]

I'd be kind of interested in the answer, too. I use FastStats (http://www.order1.net/links/affiliate.cgi?ckey=mach5&akey=b2blog&url=http%3A%2F%2Fmach5.com%2Fproducts%2Fanalyzer%2F) myself, but never really considered the affect PPC ads have on the results.

I just blogged an answer to help find out the difference.ClickTracks (http://www.clicktracks.com/new31.php) not only counts your PPC visitors, but tracks their behavior so you can compare them against the average visitor to decide if they are somehow worth more. ClickTracks is very cool, but pricy.

[pielover]

This is from FastStat's support's suggestion:

It is my understanding that most PPC configurations use the main site as a referrer, and thus cannot be detected by looking at your referring information reports.

However, they do pass variables, which you can then process using a scripting language (ASP, PHP, etc.) and then redirect the user to another page, and then to the page that they intended to visit, automatically. Thus, you will be able to view the users who visited a specific page, and then count your PPC hits. Each PPC site will have to have its own page, so that you can simply count the hits on that page, and know your general statistics for PPC hits.

I think there are other ways using the trackers that you mention. Will need to think about it. Anybody else?

[stevew]

Many thanks, pielover (good username, too)

I've been looking around a lot, and the general consensus is that all web tracking software can only ever be an approximation, because the server logs are composed of very dodgy information in the first place.
Sometimes the level of approximation is very significant.

It appears that while you can get a general view of the patterns of web traffic over time (yes, the trends ... thanks, WebTrends), any attempt at science is doomed.

However, I'd still like to think that there is a way to estimate a better degree of accuracy.

I'm probably being dense, but I'm still not sure what the following means:
It is my understanding that most PPC configurations use the main site as a referrer What do they mean by 'main site' - their site, the SE used, or ..?

Take an example where we have 1,000 referrals from SE's, and our PPC campaign generates 500 click-throughs, then this means that referrals from SE listings (ie non-PPC) are between 500 and 1,000.
At one extreme, if PPC ads (we're talking five or more different sources here) are NEVER included in the referrers, we have 1,000 referrals from listings.
But if PPC ad's are ALWAYS included, then all 500 PPC click-throughs will be included in the referrers, leaving just 500 referrals from listings.
Building a separate landing page for every PPC campaign will definitely help track the source of PPC traffic (and verify the numbers we're given by the PPC operators) but it won't clear up the confusion over whether the "Referred from Search Engines" total includes PPC or not.

And if THAT can be answered, my original question has also been answered.

[pielover]

Steve--

I just pulled up one of my log files (my ISP saves by the day, which makes this easier) and found that the "referrers" from my PPC ads are listed as such:

The images and other parts of the page list the referrer as the request from the PPC with the tracking URL:
http://www.espec.com/digest/?source=overture

Of course WebTrends or FastStats is going to ignore those because they aren't pages. But for the page, it is different!

For the page requested under referrer it lists the search page:
http://search.yahoo.com/bin/search?p=temperature+chamber

But also under cs-uri-query it lists the stripped tracking URL info:
source=overture

So, to track the tracking URLs, you need to use WebTrends to somehow record the activity in this category. That would be the next step in this investigation.

Doing this opens a whole new world of analysis. If you break it down by keyword, you can compare the success of your listings versus your ads. Perhaps getting a lot of PPC on a term you aren't ranked well will make you think that terms is doing okay, when it is actually draining your cash.

[DoubleV]

Originally posted by pielover
variables, which you can then process using a scripting language (ASP, PHP, etc.) and then redirect the user to another page, and then to the page that they intended to visit, automatically. Thus, you will be able to view the users who visited a specific page, and then count your PPC hits. Each PPC site will have to have its own page, so that you can simply count the hits on that page, and know your general statistics for PPC hits.
but if google and overture tell you how many people clicked on each particular ad,how is this different?

[stevew]

but if google and overture tell you how many people clicked on each particular ad,how is this different? 1) If you don't trust them (heaven forbid!) or you just want a second set of figures.
2) More important -- if you want to carry this information forward so that you can track the source of any enquiry/sale. After all, click-throughs are a means to an end, and you can then scrap PPC campaigns (even if they generate lots of clicks) if they don't increase your business.

[DoubleV]

1) If you don't trust them (heaven forbid!) or you just want a second set of figures.
*lol*
& thanks for the reply