Google Dance Syndrome Strikes Again, Part 2

A significant change to Google’s ranking algorithm caused some Web sites to lose top positions for some search terms. The outcry from affected site owners is unprecedented. This multipart series examines issues and questions arising from the change.

Q. My page no longer comes up tops at Google for a particular search term. Why not?

Google (like all search engines) uses an algorithm to rank Web pages it knows about. All search engines make periodic changes to their algorithms to improve the results. These changes can cause pages to rise or fall in rank. Small changes may produce little ranking differences, large changes may have a dramatic impact.

Google changed its algorithm late last month. This is obvious to any educated search observer, and Google confirms it. The change caused many to report some pages dropped in rank. These pages no longer please Google’s algorithm as much as in the past.

If your page suddenly dropped after being top-ranked for a relatively long period (at least two or three months), the page likely no longer pleases Google’s new algorithm. Running the filter test may help confirm this, at least in the short term.

Although many pages dropped in rank, others rose. Those whose pages dropped are more likely to complain than those who benefit. That’s one reason why you may hear “everyone” lost ranking. In reality, for any page that dropped, another gained.

Q. Why does running the filter test bring my site back into the top results?

My belief is Google, for the first time, is simultaneously using two significantly different algorithms. The new system has been used for many queries since the change. Some queries were handled by the old system. Increasingly more queries now appear to be processed by the new system, suggesting the old algorithm is being phased out.

Why would Google run two systems? Short answer: The new system may require much more processing power than the old one. If so, Google probably initially applied it to easy queries, such as those that didn’t involve the exclusion or “subtraction” of terms.

Why do increasingly more difficult queries now go through the new system? It could be Google was testing the new system on easier queries and planned to slowly unleash it on everything else. Alternatively, Google may have intended to run two algorithms all along but was forced to abandon the plan as site owners who lost ranking raised a furor after running the filter test to see what they consider to be old Google.

The filter test has been used by hundreds, if not thousands, of Webmasters. The queries are processor intensive. They’ve created an embarrassing situation Google never faced before: Anyone can compare what appear to be old and new results. Sometimes, new results may be better, of course. But mistakes in relevancy get the most attention. They can be used as evidence the new Google is inferior.

As a result, Google may have ultimately decided it must bring all queries into the new system, if only to plug a hole it might have never anticipated opening into its internal workings.

Google won’t confirm if it simultaneously used two algorithms. I’ve spoken with them at length about the recent changes, and they reviewed this column for accuracy.

Whether you choose to believe my speculation or accept the idea Google has employed some type of filter almost makes no difference. The result is the same. Some queries have a dramatic difference from what old Google was showing.

Q. Has Google done this to force people to buy ads?

Some believe Google dropped their pages to make them buy ads. In the short term, buying ads is the only way they can be found. For some, it may be the only long-term solution. In either case, that means more money for Google.

Plenty of advertisers lost their free top rankings. Other people never ran ads yet continue to rank well. It’s difficult to conclusively say the change was ad-driven.

Google flatly denies charges it’s trying to boost ad sales this way. The company says the algorithm change was part of its continual efforts to improve results. Google has always said there’s no connection between paying for an ad and getting listed in its free results.

There are far easier ways Google could boost ad revenue uptake without sneaky, behind-the-scene actions. That’s why I tend to believe this isn’t what’s behind the change.

Google could make the first five links on a page (rather than the first two links) paid ads for certain queries. It could also make this happen for commercial-oriented terms and say they’ve determined commercial intent of a query justifies this.

Q. Is there really no connection between ads and free listings on Google?

In terms of boosting rankings, yes. I believe this doesn’t happen on Google. Nor does Andrew Goodman in a recent article about the changes. Other serious observers also doubt this, though certainly not all. Those in the “I believe” camp feel Google would risk too much in the long term for any short-term gains.

In terms of listing support, buying ads may be helpful. Some who spend a lot on Google paid listings have reported success in getting ad reps to pass problems regarding their entirely separate, free listings along to Google’s engineers for investigation.

To a degree, this is a back door for support. Those who aren’t spending with AdWords have no such speedy solution. Google continually rejects suggestions it offers “listing support” for the paid inclusion program, saying it fears this might be seen as establishing a link between payment and free results. Google flatly denies those who advertise get better access. The company says it takes feedback from many sources, and every report is assessed for how it might impact search quality.

It’s important to note Google does provide another back door plenty of nonadvertisers use. On WebmasterWorld.com, public and private messages to “GoogleGuy,” a Google employee monitoring discussions, have been acted on.

Google reps at search engine conferences also provide assistance to those with questions.

Google also offers a front door, with published email addresses. Yes, you’ll likely get a canned response. Yet some get more personal attention. It’s critical to make the huge distinction between listing support and rank boosting. Investigating why a page may not be listed at all (as opposed to ranking well) is an appropriate activity for Google or any search engine. Boosting the rank of a particular page in return for payment without disclosure is not acceptable.

Q. Does Google have a dictionary of “money terms” it uses to decide when to filter out some Web sites?

This theory emerged as people ran the filter test and discovered for some queries, Google shows many more changes than for others. The Scroogle hit list provides a long list of examples. It reflects 24 hours’ worth of queries people tried at Scroogle to see if they declined in the new ranking algorithm. Terms with many changes are at the top of the list.

The Scroogle hit list recently showed the top 99 of 100 results in a Google search for “christmas present idea” were different under the new algorithm. That’s not entirely accurate. But overall, it’s close enough. For that query, things radically changed. The same holds true for terms such as “diet pill” and “poker gambling,” both of which can be considered highly commercial.

That’s where the money-terms idea comes from. Sites aiming to rank well for these terms may expect to make money. Some believe Google thus decided to filter out some of these sites (particularly ones showing an intent to optimize their pages for Google) and force them to buy ads.

A compelling theory. Yet many commercial terms show little change, such as “christmas time,” “books,” “sharp ringtones,” and “games.” The hit list is compiled by those checking their own terms. As you’d expect, it’s heavily skewed toward commercial queries. If a bunch of librarians entered a mass of noncommercial terms, there may be some dramatic changes in that class of queries, too.

Someone obviously tested noncommercial searches. A search for “1 2 3 4 5 6 7 8” was on Scroogle’s hit list. It came up with a score of 36 dropped pages, high enough to make you think the phrase may be in the money-list dictionary. Yet nothing about it seems commercial.

No doubt, the new algorithm appears to have heavily impacted many commercial queries. This seems more a consequence of how the new algorithm works rather than a play for certain terms. New criteria regarding how much links count; whether to count particular links; when to weigh anchor text (text in a hyperlink) more; even what’s considered spam, all probably impact commercially oriented queries more.

It’s possible Google also uses AdWords data. It wouldn’t be difficult to examine what terms attract revenue and use the data to make a list, or even to feed the new algorithm. Google won’t say whether it uses some type of list or not.

In the end, whether a predefined list of terms exists or something’s happening as a consequence of the new algorithm is moot. The result is the same: Many sites that did well in the past no longer rank so highly. Many feel targeted.

Next: more Q&A on the Google Dance Syndrome.

This column was adopted from ClickZ’s sister site SearchEngineWatch.com. A longer version of the column, which answers additional questions, is available to paid Search Engine Watch members.