Signal Strength - Perspectives on No-Follow

Posted by Eric Ward on August 10, 2010 to anchor text, link building, link signals, search rank

One of the unfortunate realities of link building and search rank is none of us really know the exact degree of impact the hundreds of off page signals have. We can read the writings of everyone from the darkest black hatter to tips from the search engine engineers themselves, but at the end of the day we cannot say with perfect clarity to what degree any given "signal" matters.

There have been countless studies and "expert interviews" where people assign a rating to certain signals. But among the top SEOs and link builders, there is plenty of disagreement. There are people I have utmost respect for as link builders who I disagree with on a signal by signal basis.

This post today will mark the beginning of a series of posts devoted to link signals. The term link signals is itself a bit confusing. When I refer to link signals, I am referring to both the pages (URLs) on which links exist, as well as the coding and attributes of individual links on those pages.

There are potentially hundreds of signals that are given off by pages and links. Some of them you are likely quite familiar with, like follow/no follow, anchor text, or link age. Others you might never have heard of, like reciprocity ratio or non-duplicated geographic inbound dispersion(yes I did make that up but it's also real). Even link color might play a role. I'm glad the search engines are sophisticated enough to better recognize when any given signal can be gamed. It's really not very hard to create a completely engineered anchor text inbound link profile that is heavily peppered with your keywords. And that engineered anchor text sends a signal to the engines that might even help improve your search rank.

For a while. Then? Boom. As the engines get smarter, your engineered signals will be discovered, and it's game over. It happened back in the day with on-page signals, and it has happened with off-page signals for years. Remember when directory signals were devalued? That was a painful moment for anyone who was foolish enough to base a linking strategy on something anyone could do.

Paid links are in the cross-hairs today. It can be very difficult for a bot to detect signals that indicate a link was paid for, and one of the bigger myths is that search engines have told marketers that buying links was was a violation of quality guidelines. A bought link may in fact have plenty of signals that indicate it can be trusted. No signal exists in a vacuum, either. You can't just look at a sites' anchor text without also looking at other signals to see if the site paying for that link can be trusted.

And this brings me to the great follow/nofollow debate. For those new to this, here's a backgrounder from Wikipedia. Put as simply as possible, if a link on a web page has the rel="nofollow" attribute, then search engines do not give that link credit that affects that page's organic search position. A real life explanation would be if I have a high ranking web page and I link to your site from it, this is supposed to cause your site to rank higher. But, if I insert rel="nofollow" into the html a href tag, this negates the potential of that link to help your site rank higher.

Here is where I think things can get confusing. If a link is nofollowed, then the engines do not give it credit and do not "hear" the signal the link sends. Nofollow is a muffler, right? Yes, but if I am running a search engine, and if my goal is to produce the best possible search results, do I really want to be at the mercy of millions of individual webmasters and page creators? Will they really all implement the nofollow attribute correctly on a link by link basis? What percentage of the billions of content creators around the world even know what nofollow means? What percentage of them are using a CMS that inserts (or doesn't) a nofollow tag automatically, and they don't even know it? And if we just look at the upper crust of web content, the absolute best of class content produced by the most brilliant minds, what is the likelihood those folks have ever heard of link attributes at all?

And this is where these truths lead me. A search engine can follow whatever the heck it wants to follow, if it feels the link will help it produce a more accurate and useful search result. No search engine dependent on an algorithm created by hundreds of genius PhD's can turn control of that algorithm over to the web masses via a hoped for perfect implementation of an arcane little known tag attribute. That's just plain silly.

Real world example. When I do a search for photosynthesis at Google, I get over 20 million results. Front and center is the Wikipedia entry. Out of 20 million possible pages, Google feels this is the best place to start, and I can't argue with them. Now, we know Wikipedia links are nofollowed (as they should be to hopefully keep that spam away). At the same time, let's think through this algorithmic logic.

- If Nofollow works perfectly and with 100% certainty, then...

- Google is displaying a page from Wikipedia as being the best page out of 20 million results, but...

- Any links on that page are of no algorithmic value to Google.

And that makes no sense whatsoever. It's like saying Albert Einstein was the smartest man in the world, inviting him to the head of the table, and then claiming every single thing he says can't be trusted.

A more likely scenario is what I believe to be the case. As I stated earlier, any search engine can choose whatever it wants to choose as the signals that produce the best result. Regardless of what we do or DON'T put in the source code, the bots have a job to do, and that job is often at odds with content publishers and web marketers.

As a link builder, I have never let the existence of any attribute impact my choice of link target. For me, it has and will ALWAYS be about relevancy and intent of the content where I am seeking the link. And scoff if you wish, but I have also seen rankings improvement even though some inbound links have been nofollowed. While I have absolutely no proof that any search engine does in fact ignore nofollow, it would seem to me that if other signals indicate they should, then they will. There job is to produce the best results, not police the links on the web. It just so happens they've ended up having to do one to accomplish the other.