How Google Panda 2 Update Hit eHow and Others

13 Flares 13 Flares ×

Earlier this year, the webmaster, online publisher and SEO communities were abuzz with news of a major Google algorithm update. Although Google claimed less 15 percent of websites were affected, legions of web publishers and search engine optimization experts took to forum, blog posts, and Google help pages to decry what they saw as unfair treatment. Whether Google’s algorithm was broken still depended on if you were one of the unfortunates.

Various web traffic measuring firms published winners and losers from the original Panda update showing that sites like Suite101.com and HubPages had been hammered while eHow and others had emerged mostly unscathed.

panda-update-googleLately, another round of complaints have come courtesy of those who escaped Panda I but were subsequently slaughtered by Panda II. The second Panda update affects even fewer websites than before, according to Google. The most notable hit this time around was eHow.

What Changed In Google Panda 2 Update?

That eHow parent Demand Media escaped mostly unscathed from the original Panda update was a head-scratcher for many. After all, if the update was to go after the so-called content mills, then shouldn’t the mother of all content mills be hit?

The quick re-update released by Google may have been in response to such criticism. It may have been that no one was more surprised than Google that SEO content generation master eHow slipped away from the original Panda update.

What changed between Panda 1 and Panda 2?

Google, of course, has been only vague about what things were changed to update its all-important search algorithm.

What many people have seized on is Google’s statement that having lots of low quality webpages on your website can hurt the high quality webpages on your website.

That’s a nice theory, but there is a major problem there. Google has admitted that they cannot algorithmically determine what is high or low quality on a single page of text. That means that Google is using other factors to determine, by proxy, what is high or low quality. What are those new factors?

One Google source mentioned that having too many ads, or too convoluted of navigation, things that deliberately put the monetization of a webpage above its usefulness could be a factor. This makes sense if that was how the first wave of Panda was implemented.

Consider that among its many sins, the one thing eHow does not do is complicate its webpages. Sure there are plenty of ads, but the content runs uninterrupted down the center of each page. Likewise, reasonable content based navigation takes prominent places on each page. The top-left placement is links to other content, for example, not an ad (which is below). Likewise, at the end of the articles are more ads, but they are Google’s own text ads, right where the company recommends placing them. The graphics are all normal and natural based on the content. There are no garish, giant graphics or other usability sins. In fact, the quality of content (which cannot be judged algorithmically, yet) aside, there are no real design issues with Demand Studios.

What Google Penalizes In Panda 2 Update

What Google may be penalizing in its second round of updates is likely to be less about what is on the page, which is what they did their best to judge in the first round of Panda, but rather what is not on the page.

Consider a site like eHow. With millions of pages of content, there is one thing that makes website like eHow different than other websites: links.

Don’t get me wrong, eHow has plenty of incoming links, it makes sure of that. And, that, is precisely what Google could target.

Consider that eHow has a dozen articles, or more, about many topics, each with a title that differs by just a few words, and each cranked out quickly and cheaply by freelancers who get a flat-rate fee for each published article. How could such webpages ever build up any incoming links? How could they ever be found in the first place if traffic was not sent there directly by Google?

The front page can’t hold every article published each day for more than a few seconds each. Writers get a flat-rate pay with no additional revenue sharing or other reason to build their own links. And eHow isn’t exactly a site that you brag about writing for.

In other words, the only links going to eHow articles are eHow links. Tons of them. Tons and tons of them.

Google has always counted incoming links from the same place as worth less than links from multiple websites. That makes sense. Internal links are like your mom saying that you’re cool. Sure, she really believes it, but that doesn’t make it a worthwhile assessment.

But sites like eHow can overcome reducing internal link worth with sheer volume. Assume that each “same site” link were valued at 1/10th of a unique site link. In that case you need just 10 of your own links to equal 1 “real link”. Even a 1/100th or 1/1,000th, eHow is one of a handful of sites on the Internet that can send that many links to each and every webpage it has, and it can do it dynamically to ensure that every page gets incoming links of some sort.

With Google’s over-reliance on the title tag to determine any webpage’s relevance, only a handful of link power is needed to push a webpage to the top of the search results when the title is an exact match to the search performed, and eHow has more than enough links for that.

Of course, ignoring internal links can be a bad move. The more times the Washington Post links to its own article, the more likely it is that webpage is a definitive source of information about the topic. The catch is, that the same article is also  likely to have several of incoming links from non-Washington Post websites. That gives Google a way to separate “good” internal links from dime-a-dozen computer generated internal links.

As we’ve seen, giving internal links a lower value cannot overcome massive internal linking. But, what if Google tweaked its algorithm to count external links, judge the relevance, and ONLY THEN count those internal links. That would make those exact match titles that eHow depends on so impotent to overcome “real” content, no matter how many internal links it throws at them.

Another possibility is that Google stopped counting links that are computer generated. Crawl the same page more than once and only count the links that haven’t changed. That would take a lot more power than the above solution, but it would ensure that only “real” links were counted. Of course, that would diminish the power of worthwhile lists of dynamic links like “Most Popular” or “Most Commented”.

Regardless of how it was actually done, what Panda II did was to change how internal links were used in determining the quality or relevance of a webpage. Having done that, eHow’s millions of pages all became members of that sad group of webpage unlinked by anyone but their own website.

Still, SEO experts and technology writers around the web insist on harping on the Panda I update news about low-quality content somewhere else on your site. They might be right, but the way Google is judging low-quality is now very different than it was just a few weeks ago.

13 Flares Twitter 7 Facebook 5 Google+ 1 Pin It Share 0 Reddit 0 StumbleUpon 0 13 Flares ×

2 thoughts on “How Google Panda 2 Update Hit eHow and Others”

  1. SMS says:

    From my tests and experiments this update is only focused on those sites has low or copied content.

    Legit sites won’t be affected

    1. WGHubris says:

      Of course, “legit,” is in the eye of the beholder. The interesting issue is that eHow was left mostly untouched by the original Panda update but hit by the second wave. The point of this particular post is to speculate on what Google may have used in its algorithm to finally hit the target.

      Thanks for your comment.

Leave a Reply

Your email address will not be published. Required fields are marked *

13 Flares Twitter 7 Facebook 5 Google+ 1 Pin It Share 0 Reddit 0 StumbleUpon 0 13 Flares ×