8 Arguments Against Sculpting PageRank™ with Nofollow
This post will explore several arguments against using nofollow for internal PageRank™ sculpting. It is based on a presentation I did at SMX Advanced. You can view my presentation slides here, however the deck is very brief since I only had about 10 minutes to present my position. The bulk of the discussion is treated in this post.
Summary of My Position
Sculpting PageRank™ with nofollow is not going away. Limited use of nofollow is recommended. However, widespread use of the technique is usually a bad idea. I recommend slowing it down.
Nofollow sculpting is a tool to raise search rankings by distributing PageRank™ flow internally. However, PageRank™ is a patented, secret algorithm designed by Google that we can never have access to. Therefore, this technique is dependent on Google and specifically targeting PageRank™ which we cannot know. It’s also dangerously granular, largely untested, not standardized across all search engines, focused only on search engines at the expense of users, and vulnerable to abuse.
Each point is covered in more detail below.
1. More Control, not Full Control
Having a mechanism at the link level to control spider behavior is a powerful tool. It has the potential to give us great control. However, without a full understanding of our sites PageRank™ (PR), we truly don’t have an accurate idea of its effects. How much PR does any single link leak off a page? How much PR do we have to work with in the first place? Without knowing, how can we best allocate our domain’s PR to the most important pages on our sites?
Rebuttal –> It’s just a control mechanism at the link level. It’s another tool.
Understood. But that’s not getting to the core of the issue. We are attempting to control the flow of PR on a domain or page, but we:
a. Have no idea how much PR we have on the domain
b. Have no idea how much PR we have on a single page
c. Have no idea how much our PR fluctuates
d. Have no idea what a link is “worth” in PR
Using the nofollow attribute to sculpt internal PageRank™ is a little like being given a very precise surgical instrument, and then forced to use it blindfolded.
2. It’s a Distraction and Masks Other Issues
Many times the deployment of nofollow masks issues such as usability, keyword dilution, and page focus. When you have nearly 1,500 links on a page, all nofollowed, there’s a problem and an opportunity to improve the user experience. (Please note I’m not picking on Zappos, who are fully aware of the issue with their old site and addressing it with the kick-ass zeta.zappos.com redesign.)
Using nofollow is also a distraction. There are many other strategies that have a stronger ROI, such as focusing on content development, link building or community. Matt Cutts states that sculpting internal PR is “a second order effect.” Make great pages and resources. Don’t waste time worrying about sculpting internal links.
3. It Introduces Management Headaches
Large sites in the enterprise are most likely to have wide adoption of nofollow. With many different departments working on different areas of the site, the additional management overhead can be significant. Imagine a situation (which is quite common) where a content team controls what appears on a page (including links), a design team controls copy but also metadata and titles, and a development team controls internal linking and metadata. With so many hands in the cookie jar, things get complicated quickly. Unless there’s a very thorough documentation procedure in place, turnover and crossover in job roles can cause issues.
Imagine sitting down to a new position and seeing several nofollowed links on a page. What’s the rule that’s in place to decide what gets followed, and what doesn’t? And what department is in charge? How are nofollow strategies preserved through site updates?
4. It’s a Band-Aid™
Fundamentally, we have the most control over crawling behaviour by creating sites architecturally sound. A very granular tool such as the nofollow attribute can aid and assist in some cases, but it’s often better to address the cause of problems within a site architecture than to slice it at the link level. Sculpting with nofollow is simply a Band-Aid™ placed on a site to fix a symptom: the cause of the symptom is poor site architecture. When you treat the symptom, you don’t address the underlying cause. Sculpting PR is never a replacement for solid IA and design.
5. What About the User?
The user experience is pretty much an afterthought in cases where nofollow is being implemented. Think of a scenario such as this one:
A site with large amounts of domain authority and ability to rank, uses that internal juice to drive PR to average quality internal pages, which in turn gets those mediocre pages ranking above better quality content in the SERPs with less authority. How does that help Google’s goal of improving the user experience?
I also fear this technique will grant even more power to authority sites on the web. The gap is already widening between small players and the large, trusted sites. By basing a strategy on the flow of PageRank™ internally, high-authority domains with more PR are more likely to benefit. This puts the balance even further onto those kinds of sites, and the rich get richer.
Rebuttal –> It works well.
Jeremy Zawodny pointed out with nofollow for blog spam, that once you introduce an economic element into the meritocratic web ecosystem, you change the ecosystem. Blog spam is worse than ever, and we don’t need nofollow for something that scripts like Akismet work so well to address. Nofollow took away the link reward of posting a comment and made links even more of a commodity that webmasters should hold onto and dole out based on uncertain thinking.
Nofollow on a site is the same principle. The economic element is PageRank™. You are given an economic alottment by Google – how much you have to spend on your pages depends on factors outside your control. Why focus on that? Why not focus on creating great resources? PR sculpting is not necessary to rank. It can be a competitive advantage but is far too powerful a tool for us to deploy without knowing more about it.
Are we letting the search engines dictate how we design our sites, and how we link internally? They’ve already put excessive control over link citations on the web and control much of how we link out. Now the same thing is happening with how we link internally. We base it on the idea that we’re working with a limited supply of ranking juice; the paranoia of nofollow on outbounds is starting to creep into internal linking. Consider the following comment:
I also link to many web sites as clarification and for definitions, but to keep Google from thinking I’m being paid to link to those sites, I’m going to add rel=”nofollow” to the links. Google might flip the other direction and start punishing me for using rel=”nofollow” so many times, but Google is already punishing me for not using it, so I’m adding rel=”nofollow” to almost all of my external links and I’ll see what happens.
It seems plausible that this kind of misguided thinking will also be applied to internal linking strategies.
6. Here Comes the Abuse
Getting this granular opens new opportunities for abuse, opportunities we haven’t even thought of yet. Here’s a scenario:
Imagine link injection – an unscrupulous person gains control of a site, inserting her own links into the pages in discrete locations (so as not to alert the webmaster). Every link on the page is nofollow’d except their own inconspicuous text links.
An additional concern is filtering by the search engines. At what point does Google (or any engine) question the implementation of nofollow? When all but 1 or 2 links on a page are followed? When 12 are nofollowed? Never? Always? There is no way to know. Google can and will change the rules at any time. Tag attributes such as nofollow are open to interpretation by each engine, and no standard exists.
Don’t become dependent on nofollow or you’ll end up regretting it when they decide to start implementing automated filtering on “aggressively nofollowed internal link profiles”. Sound plausible to you?
Rebuttal –> Matt Cutts/search engines/Google say it’s okay.
There are always good and bad implementations of any single technique. Matt Cutts may tell us a bad implementation is not okay! Do what’s right for your visitors too, not just what’s right for search engines. This ultimately gives the search engines more control over sites on the web, simply because they have the power to change their minds at any time.
Matt Cutts also stated,
The best-known use for nofollow is blog comment spam, but the mechanism is completely general.
It’s a completely general mechanism. Which means, there must be specific implementations of it that Google may not have considered. It’s only a matter of time until it’s abused and devalued.
Rebuttal –> It’s been done before nofollow existed
To me advanced SEO always balances doing what’s right for search engines with what’s right for users. Nofollow puts the balance on search engines.
7. Too Much Focus on Search Engines…
Nofollow is a technique specifically for search engines. While much of SEO involves doing things just for search engines, remember the axiom (admittedly now showing its age): create sites for users, not search engines. Other quotes relevant here contained in Google’s Webmaster Guidelines document include:
Does this help my users? Would I do this if search engines didn’t exist?
Rebuttal –> We do plenty of things just for search engines.
We do, and catering to search engines is important. However, it shouldn’t be the primary focus and it shouldn’t subtract from the target users of a site: your visitors.
Udi Manber tells us that Google’s total focus is on the user experience:
…the goal is always the same: improve the user experience. This is not the main goal, it is the only goal.
We must address the question of PR sculpting in this light. Can it artificially raise the ranking of an unimproved and relatively mediocre page at the expense of the user experience? The answer is obviously yes, but the question is rhetorical. There will always be good and bad implementations of any technique. In the case of sculpting PR, I imagine poor adoptions of it are going to be commonplace. The user experience will ultimately be sacrificed, and even if it’s in subtle ways you can bet Google will address it. That’s a believable scenario for nofollow filtering penalties.
… and Too Much Google
We’re primarily focused on Google w/ nofollow sculpting. What does that say about reliance on them for search traffic, when a specific strategy is focused on a single search engine? Last I checked SEO was about internet marketing, not Google marketing. Yes, Google dominates in a way that’s never been seen before in the history of the web. But that doesn’t mean we should be making our sites for Google.
8. There’s No Standard
There are multiple definitions for nofollow, each engine may treat the attribute differently, and there is no set standard. The official and/or original definition of nofollow is, “I don’t endorse this page,” which is inconsistent with the goals of nofollow as it’s used in PR sculpting and may or may not be consistent across all the engines.
Let’s not forget that the implementations of nofollow have changed over time:
• First, nofollow was to be used for blog comment spam
–> don’t pass reputation, don’t endorse
• Next, nofollow was recommended as a signal for paid links
–> don’t pass PageRank™ for ranking; declare sponsorships
• Then, nofollow was recommended as a PageRank™ sculpting tool
–> don’t leak PR off a page
Yahoo and Microsoft might handle NoIndex slightly differently which is little unfortunate, but everybody gets to choose how they want to handle different tags.
Inconsistencies with tags such as noindex and nofollow highlight the fact that we don’t actually know what the behavior of each search engine will be when we deploy them. As Eric Lander points out,
Links are intended to be contextual. Why then do we limit our ability to reason and just assume that the engines see a nofollow and forget all references between two linked pieces of content?
What happens to sites with wide adoption of nofollow (even reliance on it) if the rules suddenly change? It’s already morphed from a comment spam tool that prevents the passing of PageRank™, into a bot control mechanism for crawling. There are too many inconsistencies in its application, and too much variance among the major search engines.
I’m hoping my arguments here will give balance to the issue. I don’t feel I have the final answer (by any means), I’m simply looking at the issue from the other side. If we explore the issue from both sides, we’ll be able to learn far more than simply jumping on the bandwagon and trusting in the technique because others are doing it, or because Google says it’s okay.
Other links of note not already cited:
• Lee Odden interviews Google’s Adam Lasnik, and Adam makes some interesting statements about nofollow, including advice to not focus on it:
• Aaron Wall outlines information about nofollow on his membership site. If you’re reading this blog, you should be reading Aaron’s blog and his membership resources are excellent. You can find Aaron’s post on nofollow here: http://training.seobook.com
• This is an interesting and thorough treatment of the subject in an interview format. Ben Welch-bolen asked 10 search marketers their opinion of using nofollow to sculpt PR, and combined it with statements from Matt Cutts: www.searchenginemarketing.co.uk/blog/sem/pagerank-sculpting/
Please let me know if I’ve missed any worthwhile resources.
Google, Yahoo and MSN announced during our panel that they’re striving to standardize on various tag attributes (including rel=nofollow) and robots.txt directives. Interesting development! Here are the pertinent links:
Announcements about robots.txt and the Robots Exclusion Protocol (REP):
Google is providing more information about the nofollow protocol too:
I’m not sure, but I can’t help feeling the new nofollow page was partially spurred on by my presentation (since Matt Cutts and Evan Roseman of Google were both in attendance). It had probably been planned for awhile; maybe my addressing it finally pushed them into action.
Also check out this post on rel=canonical implementation.