Mike Levin SEO

Future-proof your technology-skills with Linux, Python, vim & git... and me!

PageRank is Dead - NOT!

by Mike Levin SEO & Datamaster, 03/18/2009

I was sitting in a meeting recently, and an unnamed person from another agency threw out the “fact” that PageRank is dead. This opinion was put forth originally way back in 2003 by Yahoo employee Jeremy Zawodny when Google suppressed the rampant infestation of blogs in the search results. More fuel was thrown on the fire recently with the “Aaron Wall” or “Vince” update to Google that seemed to give large brands a boost.

Well, we are over 10 years into the dominance of PageRank, and 6 years since the first proclamation of its demise, and it’s still going strong. If you build a link from a high PageRank page (say, 6) to a page that’s even a little topically related and has its title tag, URL and headline set correctly, then you stand a very good chance of deliberately moving it to the first page of Google results on that term. Of course, the less competitive the term is, the easier the control. No, what’s changed is not the PageRank algorithm, but rather the weighting of different factors.

These factors basically amount to filters that automatically penalize of special cases. Those special cases are often called outliers, because they throw off the beautiful charts you could otherwise draw from the data. If you scaled something from zero to 100 because all your data points lied in that range, but one data point was 1000, you simply drop the 1000 so that your chart looks good. You determine the landscape from the rule—not the exception.

The corollary to this is that certain factors occasionally offer an extra boost, like how long a domain has been registered under the same entity. Something along these lines could have been used for the Vince update, such as referencing sites against publicly traded companies—a good indicator of relevancy, I would imagine. It’s speculation on my part, but just because there’s a positive filter doesn’t mean that the underlying data being filtered isn’t there.

And each little edit that Google does to their relevancy algorithms throws everyone into a tizzy, saying PageRank is dead, or that there’s been a fundamental change to their scoring mechanism. It’s just not true. The sites I monitor were not terribly adversely affected by the site named “Florida” years ago that originally made this issue all the rage among SEO’s, and they weren’t affected recently by “Vince”.

Why? Because I pay meticulous attention to site hierarchy and the many indicators to Google that they are relevant to their subject-matter. Anchor text counts. It has always counted. It continues to count. What has changed is that if you optimize a widget that everyone picks up and runs on their site, say 1000 times, the anchor text of all those inbound links is identical. They all lead to the same exact landing page (usually a homepage). And they all appeared within very short order of each other. In other words, easy to identify as an outlier, and reduced in influence.

This is how certain “SEO tricks” run their course with Google and stop having their full impact. Googlebot is still crawling sites. Content is still being analyzed and indexed. And the underlying PageRank algorithms are still accounting for, I would guess, 65% to 80% of the factors of whether pages are relevant or not. As you put more words into the search box, the less PageRank plays a role, and the more on-page criteria such as title tags and headlines matter. And certain terms trigger off “Universal” search, which is in-part (and only in part) pulling from different data sources and being “combed in.”

But as anyone seeing the universal results can quickly notice, while local and video results are being given incredibly prevalent locations on the search result pages, they are clearly the radical minority of overall search results. You only see one video thumbnail once in awhile as you page through the results, and rarely see local beyond page one.

Folks advocating PageRank’s death sometimes offer as evidence that lower PageRank pages can beat higher ones. Say, a PageRank 1 page beats a PageRank 3 page. The problem with that logic is that PageRank reported by the Google toolbar isn’t very accurate, and there could be a PageRank 6 linking to that PageRank 1 page you don’t know about. PageRank doesn’t always get reported in the toolbar, but it still has its effect. Plus, if the search had 3 or more terms, there’s a pretty good chance that on-page criteria short-circuited PageRank.

What other criteria could there be? Well Google themselves admit that there’s about 100, and that’s info from years ago, so today I’m willing to bet there are several hundred, now including a diversity of feeds. Feeds now can re-crawled and re-indexed just like Web content, so the publicly available feeds from all the various social media sites suddenly become valid candidates as relevancy factors. And indeed, when a YouTube video appears in Google search results, you can see how many stars it has, so Google is looking—at least, beginning with a site they own.

So the bottom line is that after over 10 years, PageRank is getting a little creaky, and has been patched and re-patched like a piece of aging software. But it’s still very recognizable as its original form.

Don’t believe me? Get control of a PageRank 6 page that only has about 5 to 10 links on it. Launch a brand-new site. Add a page and make sure your title tag, URL and headline all match, and then make a link to it from the PageRank 6 page using the targeted term in your anchor text, and watch what happens. You’ll be spidered, indexed and included in the results inside 2 weeks. And that’s the way it should be, because if you control a PageRank 6 page, you did something to make your opinion count for something.