|
Wednesday, July 26, 2006
I know, sometimes it seems like all we post about here is Google. Aren't there any other companies? Danny's "Marcia, Marcia, Marcia" comment of years ago seems more true than ever.
To everyone's credit: we try. Ask Jeeves execs have been the keynote speaker at two recent Search Engine Strategies conferences. But there is only so far you can take this. I am sure that the room will be packed at San Jose Marriott Convention Center in three weeks when Google CEO Eric Schmidt joins Danny for a keynote conversation.
If you scan 50 search engine related posts and headlines on any given morning, the most noteworthy one is bound to be related to Big G.
To the substance of today's post. Google has been talking more about openness and disclosure of late. In line with that, they plan to disclose how many clicks you haven't been charged for on a daily basis... those that have been filtered out and deemed "invalid." That will go a long way towards clearing up people's confusion about the process.
However, sometimes I hear about very basic discrepancies that would tend to indicate that Google is not catching all the *worthless* clicks or clicks that any rational assessment would call "non-clicks". We know search engines don't consider a page very relevant over time if someone clicks through on a listing and hits the back button *immediately*. So why are we being charged for many of those "clicks"?
Indeed, it's been industry knowledge from the beginning that many visits from any source -- and particularly from paid search links -- result in "short visits" of fewer than five seconds. If Google filters out 10% of clicks as invalid but a typical proportion of "short visits" is 50%, then there is still a lot of waste built into the whole idea of paying for a click. Anyone who studies basic web analytics has known about these high short-visit proportions for some time.
So I'm often asked questions like "our analytics say we only got 100 clicks, and Google's claiming 180." That kind of discrepancy is often troubling, but all it could mean is that your analytics software is not interested in counting (or unable to count due to technical limitations) "very, very short visits" of less than two seconds. And if these visits are *that* short, then they must often be click fraud. At the very least, they're utterly worthless clicks.
Now that Google Analytics is installed on many advertiser accounts, these "ultra-quick exits" are plain for all to see. We know that Google knows about them, and Google knows that we know that they know.
So we'd expect some in the analytics and advertising communities to continue questioning the practice of charging advertisers for most of those "fast-exit" clicks. As far as I can tell, Google and the other vendors do not call most of these fraudulent or invalid, and so the very notion of a conversion rate from "clicks" is skewed. Advertisers with a 4% conversion rate from "click" to "sale" may be disappointed in that ratio. But if you were able to pin down only those clicks which met a commonly-acceptable *advertiser* definition of a click, and begin counting conversion rates from there, I would guess that typical advertiser's conversion rate would look more like 8%. Put on a happy face. It's not you, it's them.
Posted by
Andrew Goodman
View Posts by Category |
|
|
