Google, Defender of the Faith
Mysterious rumblings and clankings from within Google’s Mountain View, CA, stronghold suggest it, and the wailing and gnashing of teeth in search engine Web forums appear to confirm it: The most popular search engine in the world seems to have made some important adjustments to the way it ranks Web pages. This matters because those ranking formulae determine whether a Web page appears in an organic search results page. Appearing in the top five slots—“above the fold,” in search parlance– can make a big difference to a site’s Web traffic, since many searchers don’t venture too far below that first screenful of search results.
How Google determines a page rank depends on many factors, but one of the most important is the quantity and quality of the incoming links it contains. Google relies more heavily than the other search engines on the number of sites linking to a page as a measure of how useful users find that page, and therefore how high it should rank in a search for the terms contained on that page.
That was fine in the early days of search, when a link represented a vote of confidence in the value of the page being linked to. But as search has become more marketing-driven, a substantial cottage industry has evolved in providing those incoming links for financial benefit. In other words, some proportion of the links on Web pages today are placed there not because the content is valuable—the basis for Google’s PageRank measurement—but purely to raise that page’s search position.
One of these techniques involved so-called link farms, chains of Websites that were intentionally built merely to link to a Web page and elevate that page’s organic Google ranking. That trick was impaired by the last big change in Google’s algorithm in November 2003; it’s still referred to as the “Florida update,” the way survivors spoke of the Great Chicago Fire and the San Francisco earthquake.
Extrapolated out over time, this trend could undermine Google’s usefulness as a search engine. And in a company that, according to its most recent quarterly statement, gets 95% of its revenue from paid advertising including search ads, anything that impairs the basic value of its search function is going to demand attention—and get it.
“Google is the number-one search engine in the world because we the users agree that it is,” says Jim Hedger, consultant and evangelist with Canadian search engine optimization (SEO) firm StepForth. “When our brains say ‘search,’ our muscle memory types ‘Google.’ Alta Vista used to be the number-one search engine, but then people started losing faith in the accuracy of Alta Vista’s organic search results and turned to Google. If people start losing faith in Google’s organic results, Google is afraid they’ll go and find an alternative search engine. And if they lose faith in the organic results, Google is afraid they’ll lose faith in the paid results. And that’s Google’s worst nightmare.”
And Google now seems to be in the process of tending to those link problems within its page-ranking formula, Hedger says. “Seems” is the operative word here. It’s impossible to say for sure, since only Google search engineers know for certain what changes are being implemented. Google, Yahoo!, and the other major search engines are very close-mouthed about their algorithms, knowing that talking too much about what they do will encourage programmers to find way to manipulate their processes.
So observers are forced to rely on outward signs to hypothesize about what’s going on in the “black box” of page rankings. And those signs include the disappearance of a large number of Web pages that were “aggressive,” to use Hedger’s phrase, about using links to bolster their Google rankings. “These companies have been penalized,” he says. “They’re not coming up in the top ten search results anymore.” With eight billion Web sites indexed by Google, it’s hard to say that what seems to be happening in one small area is happening around the index, he admits. “But we know that they’re not where the average Web surfer would have found them in the ranks two weeks ago.” The consensus among SEO experts is that a large ranking shakeup is taking place.
Among those companies that appear to have had their Google ranks cut from under them were several SEO firms named in a March Wired article that discussed how easily SEO can manipulate Google rankings via links. Hedger believes that article was a “call to action” for Google’s engineers to start scrutinizing page links more closely.
And how are they doing that? Again, observers outside Google’s corporate walls are left to guess based on what they observe. But Hedger thinks he’s found some important clues in a patent application filed in late March by Google, covering a large swath of ranking analysis techniques. Primarily, these have to do with examining “historic data” for both the page being ranked and the pages linking to it. In the course of its indexing of the Web, Google collects a lot of data about pages, sites, links and URLs. It knows how long a page has been online, assuming that it became aware of the page fairly rapidly. Therefore, it knows how long pages linked to that page have been online. It knows how often links get clicked, which computers click them, and where the click activity is coming from.
The evidence Hedger says he finds in the 70-page patent indicates that Google can, and possibly is, using this historic data to separate valuable organic page links from those that are placed merely to raise Google rankings—“link spam”. For example, the search engine seems to be willing to evaluate how pages linked to each other change over periods of time, either seasonally or as a result of events. Do the linked pages both change, or only one?
Google can also monitor the link growth rate on a page, perhaps assuming that a spate of fresh links indicates possible spam. It is able to track the number of searches for keyword phrases used in links, and the number of times users click on results produced by entering those keyword phrases used in links. It can also watch look for variations in the anchor text used to phrase links on a page; if the text doesn’t vary, it might be a sign of branding or spam. And it can tell when that anchor text in a link is not relevant to the page being linked. As Hedger says, “A link using ‘blue widgets’ as its anchor text should link to a document directly associated with blue widgets.”
Hedger points out that just because Google’s patent filing says it can use these measures, among many others, doesn’t mean that it is in fact using them now. But taken with the apparent shifts in many prominent Web sites’ Google ranking in the past few weeks, the chances are strong that the company is scrutinizing the linking component of its ranking algorithm harder than ever before.
And that means that responsible SEO should include that same level of link scrutiny, because even if Google isn’t wielding these link-history tools now, the company will inevitably bring them out if it feels its value to the user is being undermined by questionable linking gambits.
“Basically, if your Website has links on it that stem from ‘bad neighborhoods’ you have to assume that the histories of those neighborhoods will affect your ranking on Google,” Hedger says. “The big key here is relevancy. Webmasters and the outside SEOs they employ have to be more diligent about checking backlinks for relevance.”
The organic search rankings are Google’s big loss leader, Hedger points out. “They’re not getting any money from them, but they are getting credibility and user faith,” he says. “All this industry is built ultimately on faith: faith that the results produced are relevant.”
