Avinash Kaushik has a very interesting article on measuring the value of “upper funnel” keywords.
The basic idea is that different types of keywords get you different kinds of traffic, and that it doesn’t make sense to evaluate all your keywords on the same basis — i.e., first-visit conversions.
So if you’re not going to evaluate keywords (that you’re paying good money for!) on the basis of conversions, what do you do?
Kaushik says you put keywords through a funnel analysis. Some of them introduce people to your company / website, but they’re not likely to convert. Measure them by bounce rate.
Others relate to category or brand. You measure them by time on site or return visits.
Finally you have your “conversion” keywords, which you measure by sales.
Okay, this sounds reasonable, but it sounds a little like the “branding” pitches you get from people who sell display ads. “No, these ads don’t convert off clicks, but they increase traffic to your site, brand-related searches, etc.”
Which is all well and good if you can prove it!
That’s where you get into the tricky analytics portions of the discussion, which involves “multi touch attribution analysis.” What’s that, you ask? It’s “the art … of measuring truly pan session customer behavior.”
Lovely. Can I do that with Google Analytics? (Maybe. See comment 16 in the post I link to above.)
Oh, but I said I was going to talk about landing pages.
If this theory really works, it argues for different landing page strategies for different groups of keywords.
Many people design landing pages for “conversion only.” Drive visitors straight to your cart and don’t give them any options to do anything else!
(There’s a smarter, middle ground that gives them some other options, like going to your home page, but that’s for another day.)
However, if I have a pile of keywords that are designed to get people interested in my content, I obviously don’t want to send them to a page that says “buy my widget” and nothing else.
So Kaushik’s “funnel” approach to keywords leads to a strategy that involves several different groups of keywords going to several different types of landing pages — and a headache of a time tracking it all in some fancy pants analytics tool.