PageRank Sculpting

People view PageRank through different lenses, often shaping their SEO strategy around it. Some compare PageRank to a "random surfer" model, where it represents the probability that a random user clicking on links lands on a specific page. Others envision the web as a network matrix, where the value at position (i, j) indicates the presence of a link from page i to page j. In this scenario, PageRank corresponds to the principal eigenvector of that normalized link matrix, making it a main element in SEO strategy.

The most common way to visualize PageRank is as a flow occurring between documents through outbound links. I'm sharing an image from one of the original PageRank papers:

In the image above, the document in the lower-left corner possesses "nine units of PageRank" and three outbound links. As a result, the PageRank distribution along each outbound link is nine divided by three, equating to three units of PageRank per link.

However, this basic model has its limitations. Consider the scenario where a loop exists

No PageRank would ever escape from the loop, and as more PageRank continues to flow into the loop, the PageRank within that loop would eventually reach an infinite level. Infinite PageRank isn’t particularly useful, so Larry and Sergey introduced a decay factor—essentially, you can think of it as 10-15% of the PageRank on any given page vanishing before the remaining PageRank is distributed through the outbound links. In the random surfer model, this decay factor represents the random surfer getting bored and choosing to navigate to an entirely different page. This reset vector can be used for interesting applications, like personalization, but that's beyond the scope of our discussion.

Now, let's discuss the rel="nofollow" attribute. Nofollow is a method (introduced in 2005 and supported by multiple search engines) that allows you to tag a link to indicate to search engines, “I can’t or don’t want to endorse this link.” In Google, nofollow links do not transmit PageRank and do not carry anchor text [*].

So, what happens if you have a page with “ten units of PageRank” and ten outbound links, five of which are marked with nofollow? Let’s set aside the decay factor to focus on the main issue. Initially, the five links without nofollow would each pass two units of PageRank (meaning the nofollowed links didn’t count toward the denominator when dividing PageRank by the number of outbound links). However, more than a year ago, Google altered the way PageRank flows, so now the five links without nofollow each transmit one unit of PageRank.

Why did Google modify how it counts these links?

One reason is that some teams focused on crawling, indexing, and quality noticed that certain websites were trying to manipulate the flow of PageRank within their sites. However, these attempts often resulted in excluding sections of their sites that contained valuable information, such as user forums.

Does this mean “PageRank sculpting” (attempting to control PageRank flow within your site using nofollow, for example) is a bad strategy?

I wouldn’t advise it, as it’s not the most efficient use of your PageRank. Generally, it’s better to allow PageRank to flow naturally throughout your site. The concept of “PageRank sculpting” has always been a secondary or tertiary recommendation. The primary focus should be on creating excellent content that naturally attracts links and designing a site architecture that is both user-friendly and crawlable for search engines.

For instance, ensuring that users (and bots) can easily navigate to the pages on your site by following links has a much greater impact than trying to sculpt PageRank. If you run an e-commerce site, a good example of proper site architecture would be placing products prominently on your homepage rather than hiding them deep within your site, requiring multiple clicks to access.

There might be a very small number of pages (like links to a shopping cart or a login page) where adding nofollow makes sense because those pages are user-specific and not particularly useful in search results. But overall, I wouldn’t recommend PageRank sculpting.

Why are you sharing this information now?

For a couple of reasons. Initially, we assumed that site owners or those running experiments would notice the change, but they didn’t. In hindsight, we’ve made other, more significant changes to how we evaluate links that also went unnoticed, so this shouldn’t have been a surprise. We began providing guidance that PageRank sculpting isn’t the most productive use of time.

When we added a help page about nofollow to our documentation, we stated that “a solid information architecture—intuitive navigation, user- and search-engine-friendly URLs, and so on—is likely to be a far more effective use of resources than focusing on crawl prioritization via nofollowed links.” In a recent webmaster video, I mentioned that “a better, more effective form of PageRank sculpting is choosing (for example) which elements to link to from your homepage.”

During a site review session at Google I/O, I was even more explicit: “My short answer is no. In general, when you’re linking within your site, don’t use nofollow. Just go ahead and link to whatever content.” But at SMX Advanced 2009, someone asked the question directly, and it seemed like a good time to clarify this point. Again, it’s not something most site owners need to worry about, but I wanted to inform the more advanced SEOs.

If I run a blog and add the nofollow attribute to links left by my commenters, doesn’t that mean less PageRank flows within my site?

If you think about it, that’s how PageRank worked even before the nofollow attribute was introduced.

Doesn’t this encourage me to link out less? Should I disable comments on my blog?

I wouldn’t recommend disabling comments to try to “hoard” your PageRank. Just as Google trusts sites less when they link to spammy sites or bad neighborhoods, parts of our system encourage linking to reputable sites.

If Google changed its algorithm for counting outlinks from a page once, could it change again? I really like the idea of sculpting my internal PageRank.

While we can’t guarantee that our algorithms will never change, we do not anticipate this aspect changing again. If it does, I’ll try to keep you informed.

How do you use nofollow on your internal links on your personal website?

I generally let PageRank flow freely throughout my site, and I’d recommend you do the same. I don’t use nofollow on my category or archive pages. The only place I deliberately add nofollow is on the link to my feed, as it’s not particularly useful to have RSS/Atom feeds appear in web search results. Even that isn’t strictly necessary, because Google and other search engines are good at distinguishing feeds from regular web pages.

[*] Nofollow links definitely don’t pass PageRank. Over the years, I’ve encountered a few edge cases where a nofollow link did pass anchor text, usually due to bugs in indexing that we later fixed. The key takeaway is that nofollow links don’t help sites rank higher in Google’s search results.