Need PPC Tracking Advice. Help with Analytics?

Posted By: onlineguy ()
Posted On: 2007-Feb-27 18:17

I posted this in another board but haven't had any answers.

Due to Google's recent changes, we really want to track our campaigns better. In the past, we never really used many tracking methods. We only used our Web stats package and some tracking URL's, which are very limited. This week we will be implementing a new tracking system.

We are now testing out Google Analytics and others.

We want to track the "broad keyword" that people search for. For example, if we bid on the term, 'widget', we want to know if the person who clicked on our Ad searched for 'big blue widget' or 'small red widget'.

Can this be done with Analytics? How can this be done?

Thanks in advance everyone!



See also: http://www.searchengineforums.com/apps/searchengine.forums/action::thread/forum::ppc/thread::1172499796/

[ Message was edited by: JimBot 10/15/2008 12:10 pm ]




Posted By: flyingrose (Staff)
Posted On: 2007-Feb-28 00:30

I asked Mike who created the program Hit-Tail to answer in the other thread.

There is no easy way to do this as far as I know. You can dig most of the actual searches out of your log files; however, some sites strip the actual search terms and others return long threads of junk.

I know what you want to know is what words triggered your broad matched keywords so you can control this behavior. I suspect the spending problems we're seeing are primarily bad sources of traffic and incorrect expanded broad matching.

If that information is available in Google Analytics I haven't found it yet. The "Keyword Considerations" feature would be the most likely place but it offers so few suggestions that it must be really heavily filtered.

If Google would give us a way to opt out of those two "problems" our spending would be controllable and stable again.


Posted By: onlineguy ()
Posted On: 2007-Feb-28 14:17

We found this blog article which mentions a "work-around" using the filters that are available in Analytics.

[link]

Has anyone tried this before?

Thanks again!


Posted By: flyingrose (Staff)
Posted On: 2007-Feb-28 23:01

Thanks for the find. I'll see if I can make the time to add the filters to a couple of accounts and see if that does it. When I do I'll report back here. If it works advertisers may be able to control spending with ever growing numbers of negative keywords as an alternative to being forced to pause or delete broad matched keyword phrases.

I forwarded the link to Mike Levin. Since he created the HitTail program he has more knowledge in that area than most. Hopefully he'll be able to provide us with some additional input.



Posted By: Mike-Levin.com ()
Posted On: 2007-Mar-02 19:27

Uh yeah, so in the end, you really only can have as much data as a regular logfile can contain. It's more than people think, and storing it all is usually overkill. So, G/A is making two important decisions:

1. Precisely what to store (cookie info? User agent? etc.) and how long to store it.

2. How much to make easily accessible to the users.

If Google were to take the critical last step of allowing you complete API-like access to that log file data, you could write apps that could auto-optimize a PPC campaign. Just add your own personal alg's.

This sounds something like what Joe SEM is doing in the blog post Rose referred me to. And it's probably very possible, and there may even be bid management software that's doing this today--I really don't know (but I should).

So, why doesn't Google expose this data more easily to the average user? Two likely reasons. 1) a favor to 3rd party bid management software developers, or 2) they're holding back until they're ready to reveal part of a possible end game, which is auto-optimizing AdWords, where you just set the spending level and say "Go!" No keyword lists ever need to be figured out, maintained or optimized in this scenario. You just trust Google to auto-optimize your campaigns. It would required an analysis of click-thru and conversion of course. But I expect it's coming.


Posted By: Mike-Levin.com ()
Posted On: 2007-Mar-04 00:26

Oh yeah, I didn't directly answer the question. On the topic of sorting out what traffic came from where...

1. Anything from the AdSense network is recorded as such in your logfile data as being from Google Syndication, but you lose the keyword data that brought people to those pages.

2. Anything from regular AdWords gets recorded in your logfile data as if it were a natural search hit. So the only clear way to sort it out (from your end) is to funnel all your campaigns through landing pages that specifically record the hit, and are findable in no other way. That usually means redirecting links and a home-made recording program. There may be 3rd party products that help--maybe those programs that "shorten" your URLs.

Yes, that's a bad situation. I think if you're a high-end client with a Google sales rep, the'll add a CLID (click ID) parameter to your AdWord clicks, but you have to be pretty big customer (I think).

And as far as sorting broad match hits from non-broad hits--I know of absolutely no way other than separate campaigns with separate tracking and redirecting landing URLs.


Posted By: flyingrose (Staff)
Posted On: 2007-Mar-04 01:14

Thank you for that detailed information. My muddy understanding is starting to clear to a dingy water now. I can add that any advertiser can add additional parameters (manually) to their keywords.

When I used IndexTools I was also able to manually differentiate traffic and sales driven by the broad, phrase, and exact match versions of keywords by adding a different keyword parameter to each one, for example:

kw=word+word2+type where type can be the words broad, phrase, or exact or an abbreviation of those

I actually measured the difference and that is why I know how much traffic and sales an advertiser is likely to lose if they delete broad matched keywords.

Note that I have not done this since Google Analytics (GA) became available and auto-tags keywords so I do not know for sure how GA will display or break out such designations. Every analytics program varies in how it works.