Great post. I love clever uses of filters. But there *may* be a minor oversight.

"But if you type in the expression ^cheese$, then you'll only see the 'cheese' keyword exactly."

To match 'cheese' exactly, you'd probably want something like "^cheese -" without the quotes. Since the product name will be appended to the keywords with a space and a dash. (Or does the dash need to be escaped?)

@John:

You're absolutely right! This is what happens when I try to think on a Friday afternoon. Thanks for catching this - I'll edit the post to reflect this. The dash doesn't need to be escaped.

Hey Shawn,

I sat here thinking some about your post . . .

I like that your solution puts the Item Name and the Keywords in the same field. That has some good benefits when looking at the data. But,

What about using a filter to put the Item Name in, let's say, Java Enabled? Then you could go to the Keyword report, find the keyword phrase that you're interested in, and segment by Java Enabled, and see a list of all Items purchased for that keyword.

This would be sort of the opposite of what GA provides by default. Instead of selecting a product and segmenting by keywords, you would be selecting a keyword and segmenting by product.

Any thoughts, or known problems doing it this way?

@John:

The only drawback that I can see without testing is that you'd lose your Java report (which probably isn't a huge deal - especially if you are using a separate profile). I also know that occasionally mixing ecommerce data and your other fields can be disastrous because of the ways that data is collected -- so you'd definitely want to test this first. It's also possible that this Java Enabled field is boolean (yes or no) - although I doubt it.

That being said, I think there is definitely value in using the Java Enabled? field and it's absolutely worth testing - especially if it allows people to use the Segment pulldown. Ultimately, both methods would do the same thing - and I'd love to hear if you test yours how it works out.

Thanks for the idea!

@John: I've used your approach, but stored the data in user-defined. I've only got a day of data, so it's a bit early to be sure, but I seem to be getting ecommerce data. (And I have experienced disasters and loss of ecommerce data when trying to manipulate campaign variables using filters in the past.)

@Shawn: I've also used the version you suggested in the original post and even with one day's data it's proved to be very useful. A revelation in fact. Lots of great insights into the function of brand search in the final, purchasing, visit.

Thank you.

Now I'll try to combine both in one profile...

@Tim: I'm glad the method is proving useful for you.

Before you use the User Defined vairable, you may wish to consider using Michael's Exact Keyword Tracking tool. If you're doing any kind of pay-per-click advertising, you may find it more useful than product-keyword matching. We've found it to be an absolute goldmine for negative keyword research and new keyword discovery.

Really, it all comes down to priority. You could use the exact keyword method in one profile,and the product/keyword match in another. I'd love to hear how the User Defined method works out for you.