Tuesday, December 02, 2003
"We're Always Trying to Improve the Index for the User": Google
The problem with rampant speculation is that it's usually at least as damaging as the behind-the-scenes shenanigans you're trying to speculate on. Recent armchair attempts to explain the master plan behind Google's recent index update, mine included, are no exception.
So, lest I be accused of being the Oliver Stone of search, after talking with Peter Norvig, Google's Director of Quality, I'd like to clarify a few things.
I've always known and believed that there was no relationship between Google's advertising program and its index results, absolutely none. After working so closely with so many advertisers, it would have been pretty obvious if we'd been getting some sort of positive "spillover" effect. Many of us weren't suggesting this sort of relationship, but it *wasn't* out of line to make the point that a significant reshuffle at this time of year does make many non-advertisers aware of the fact that they might have become too dependent on free listings. Google doesn't have to foster or maintain a relationship between the right and left-hand sides of the search results page to benefit from the fact that both sides are in constant flux.
[Disclaimer, if that's even the right word: my company benefits, too, since we help people figure out how to make their dollars go farther on the right-hand side of said page.]
Google's not unaware of this. A closing comment from spokesperson Nate Tyler seemed to contain something of a pointed message in this regard: "People need to be aware that the Google index was not designed to be a predictable way for companies to get traffic, although, of course, if you type Amazon, you're pretty sure to see Amazon.com up at the top, since that's clearly the most relevant result." Unsurprisingly, Google would rather have you as an advertiser than not, and if the threat of unstable, unpredictable index rankings for private-sector actors is enough to convince more of them to finally invest dollars in AdWords, then so be it.
One area that I did exaggerate a bit in my article, but again, not without some good reason, is the fact that Google can certainly collect information about the financial value, to Google, of certain search queries as those queries are monetized through the ad program. I might have mis-guessed as to how such data might be used -- and I certainly wouldn't want to suggest any kind of systematic relationship between ads and index -- but it's certainly the case that Google is at least *in possession of* information telling them which queries are commercial, and which aren't.
But that's neither here nor there.
According to Norvig, Google is "always making changes to its index, and it measures the quality of results before and after." One explanation for the current hue and cry, in Norvig's view, is simply that "Google went for a period of several months with no major changes, and some webmasters got complacent about their search rankings to the point where they felt deserving of them."
One point to make is that changes to the index don't always affect all queries equally. In rolling out product improvements like showing results with "stems" and "plurals," some queries are affected and others aren't.
The most recent enhancement, says Norvig, can be boiled down to "attempts to give the correct value to a page." This is what caused problems for so many sites who had managed to climb high in the results -- higher than their sites warranted -- by exploiting search optimization tactics. In short, in large part, this was in part your run-of-the-mill anti-spam re-ranking, but also, Google may have begun down the path of incorporating new cues to a site's quality or relevance to make the results that much more useful to the public.
"We used to look at just links and keywords, but now we're incorporating a lot of other stuff... looking for more and more signals and types of information on a page that attempts to determine or read a 'real meaning' or what a page is trying to provide," continued Norvig.
He acknowledged that some of my speculation, the part where I suggested that Google was making more effort to discern the "type" of information on a page (resource/discussion/information, store/affiliate, company, etc.) "was heading down the right path." Norvig even went so far as to agree that the type of thing Google "might" do would be to look for information such as "how long a company has been established, what kind of information is it showing to the site visitor, etc." It's safe to say in such a context that those traditional bastions of SEO, the hastily-assembled "microsite," would have trouble cracking a top ten listing under this type of formula. But wasn't PageRank supposed to be immune to that junk anyway? Is Google quietly admitting that they've got to layer more and more tests of quality into their algorithm because they're powerless to stop the growth of link farms and superfluous reciprocal linking?
And although I'm satisfied with Google's ongoing efforts to achieve higher quality, at this point, it looks as if the quality of listings is more predictable on non-commercial queries.
Because, after listening carefully to every possible factor that Google might take into account in judging quality and relevancy, when I type "fruit basket" into the search box, I'm still confused.
This is the top-ranked site on that query:
The #2 site is no work of art, either:
The #4 result for "fruit basket companies" is a bit of spam that should be caught easily:
One thing is clear. This won't be the last Google Dance. The next one can't come too soon for many webmasters.
View Posts by Category
Andrew's book, Winning Results With Google AdWords, (McGraw-Hill, 2nd ed.), is still helping tens of thousands of advertisers cut through the noise and set a solid course for campaign ROI.
And for a glowing review of the pioneering 1st ed. of the book, check out this review, by none other than Google's Matt Cutts.
Posts from 2002 to 2010