|
Monday, May 29, 2006
Since Day One of v 2.0 of Adwords (Feb. 2002), Google has strongly signaled that it is willing to trade off some revenue for long-term sustainability in its ad program. It's willing to spend cash in order to remove "unpleasant" or irrelevant ads from the system. The definition of poor quality ads, now as then, is probably a combination of user responses and company values.
These principles have not always been easy to enforce, however. Since introduction of a Quality Score concept in August (or so)*, Google has essentially redoubled its efforts to fine-tune the system so that it can understand exactly how much revenue it is forgoing when it enforces certain rules, and favors certain kinds of ads over others (regardless of bid or even CTR).
Last week (or so)*, advertisers became aware - through an update to the terms and conditions - that Google has begun spidering every landing page in the AdWords system to gain added cues for its quality scoring process. That raises key questions, of course. Weren't they doing this already? If not, what were they doing? Rumors abounded that they used human review, that they used information from their existing (index) crawls, and of course, that they were doing this crawling all along without telling us, though this latter is denied by those who watch out for crawlers.
Regardless, it is an improvement to have the rules made slightly more explicit. Post pages that don't give adequate access to the crawler - or adequate keyword cues - and you risk facing the wrath of the quality scoring algorithm. It's less of a worry as much if you have an established account - it's new accounts that face the toughest tests with the predictive aspect of the algorithm, intended to weed out specific types of violators, experimenters, and ham-fisted copywriters.
A final takeaway - and since I'm in London right now, that word is making me hungry: more so than when editorial policies were enforced by people, the quality score regime allows Google to not only penalize certain types of ad quickly and painlessly, it allows them to test and model the response in the system, and even the impact on the company's revenues. If there is a particular genre of ad that Google and its users don't like, Google can assess with some precision how much cash they're going to have to spend to buy quality for their users. Now Google even knows the cost to shareholders of setting up sanctions against various evils.
* - Explanation of "Or so": The Quality Scoring system was announced in August 2005, but aspects of it were likely in use as early as January 2005. Spidering landing pages explicitly began with the new TOC last week, but Google's been checking landing page availability, and scanning pages and whole sites in order to make keyword suggestions, for over a year - that's the little wait with the colorful animated balls when you upload a new ad, new keywords, etc. Anyway, now that we've all agreed to the new TOC "through continued use" of the AdWords system (without explicitly being asked to click a checkbox), the question of who told us what, when, is now relegated to yesterday's news. This uncertainty about the timing of certain initiatives (they happen before they're announced, sometimes) underscores both Google's secrecy and their impulse to experiment and tinker early and often.
Posted by
Andrew Goodman
View Posts by Category |
|
|
