A few weeks ago, there was a lively exchange on Search Engine Land about using the “nofollow” link attribute to sculpt PageRank. Shari Thurow, in her article You’d Be Wise To “NoFollow” This Dubious SEO Advice, essentially railed on SEO practitioners for employing this practice, which respected expert Stephan Spencer describes and advocates in his article Sculpting Your PageRank For Maximum SEO Impact.
If you do not believe that a page’s content is important, then don’t link to it. Better yet, remove the content. If you believe a web page’s content is important, then link to it and do it in a way that makes sense to your end users, your site’s visitors. I think it is very odd to put a nofollow attribute on pages within your own site. Essentially, you are saying that you cannot validate your own content—you advocate giving users one information architecture and search engines a different one?
Shari’s comments regarding the use of nofollow seem to imply some sort of bait and switch tactic that would not only fly in the face of search engines, but would be deceitful in some way to site visitors. So many people have cited Matt Cutts’ position that there is no problem with this practice that I won’t bother citing more. However, for those fearful of employing the practice, Matt indicated that employing such practice in no way even serves as a red flag to Google. Secondly, how could such a practice be deceitful in some way to site visitors? When the visitor is on the site, they have no idea which links have the nofollow attribute; they can go anywhere the navigation allows.
While it would be great if every page had the same high value to search engines and site visitors alike, that’s simply not reality for the vast majority of sites—even if it has been optimized for human usability. There are many pages that have real value to site visitors but marginal value to site owners in terms of PageRank or being included in search engine indices.
So what links should you nofollow? Stephan Spencer cites many examples:
While many of these pages may be of low value, don’t automatically nofollow links to all of these pages. Some of the pages Spencer identifies above may be sending a lot of traffic to your site through natural search. These pages may have been optimized to address specific keywords. For instance, we will often incorporate a significant amount of copy on the “about us” or “contact us” pages, both for visitors and to satisfy a desired search phrase. It’s not gaming the system; it’s relevant and appropriate information we want and need on client sites for site visitors, and we choose to use the “about us” or “contact us” pages to convey that information.
In addition to content-less pages, I recommend looking at your analytics. Isolate the landing pages coming via natural search and look at their relative popularity in terms the other natural-search landing pages and in terms of bounce rate. Then compare that list to the pages in your total site architecture.
You should evaluate those pages with little to no natural-search landing page activity and those with extremely high bounce rates. If the page has little or no natural-search landing page activity, you should determine whether you need to revise the content and other factors associated with the page to increase the likelihood of natural search click-through. However, if the page is important to site visitors, and if the content and all other factors are how you want them, you may want to nofollow links to this page in order to concentrate PageRank elsewhere. Assuming the page is indexed and has associated PageRank, the page is doing very little for you in terms of search anyway.
There also may be internal landing pages with extremely high bounce rates. If 98 percent of natural-search visitors landing on a given page leave right away, you’ll want to evaluate these pages in the same way and make appropriate changes, either in terms of content and other factors or in terms of placing nofollow on links to such pages.
The usability purist would argue that any page important to a visitor should also be important for search, and that your architecture and linking structure, if engineered correctly, should remove any need for nofollow in the first place and appropriately weight PageRank throughout the site. That utopian thinking doesn’t mesh with reality. Here’s just one example.
In the B2B world, case studies represent high-value information for prospects. Accordingly, a B2B site owner’s architecture and linking reflect a desire to drive site visitors the “home page” of the case studies section. The prospect wants to see a case study relevant to her issues. The site owner wants to present numerous and diverse case studies to help ensure the visitor finds a case study relevant to her. As a result, the site owner may have 20 case studies linked from the “home page” of the case studies section, each of which receives a portion of the total PageRank. If these case studies don’t vary all that much (at least in terms of unique and probable search phrases searchers would reasonably use), the site owner would likely be better off placing nofollow on links to many of the case studies, concentrating PageRank in other pages. While this may be a good practice, you’ll again first want to check your analytics to ensure the links you consider for nofollow aren’t linking to pages currently sending a lot of natural search traffic.
So go ahead, employ the practice of nofollow, but do it smartly. And being sure to consult your analytics as part of the process.
2 thoughts on “Choosing Which Links to NoFollow: A Practical Approach”
Search engines are getting increasingly selective about what links are actually related to your site. When building links, it is import to remember to stick with sites or blogs that have related content to your own website. If you are linked to unrelated websites, those search engines will discount your website and it will put you lower on the PageRank instead of moving you up like you want it to.
Galen, I agree with you completely on this subject. It makes absolutely no sense that a site using nofollows is showing a different architecture to users – who aren’t crawling your site like a spider. I think they’re well used for links to a login page or sometimes even comments. You’ve said it best when you gave the example of it not meshing in the real world.