Google's PageRank Algorithm: Explained and Tested

Article stats:
Linking websites 89
Backlinks 137
InLink Rank 38.0
Data from: online backlink analysis tool.
The Vice and Virtue of PageRank

Let me tell you a story. In the beginning, there was PageRank, and it made Google the most powerful search engine.

Everything was great till PageRank stayed the secret sauce of Google's ranking mechanism and was talked about just in the research papers and technology pages.

However, once Google decided to make PageRank scores visible, it unleashed a flood of optimization strategies that were not exactly benign.

Let's look into what PageRank was, what PageRank is, and decide whether it was a blessing or a disaster for the Web.

Definition of PageRank

Google PageRank article by SEO PowerSuite

PageRank algorithm (or PR in short) is a system for ranking webpages developed by Google founders Larry Page and Sergey Brin at Stanford University in the late ‘90s. It serves as a vote of trust given to a page, meaning that the more external resources link to that page, the more valuable information it has for users. PageRank is used to give each page a relative score of importance and authority by evaluating the quality and quantity of its links.

Here's how it works. Each link from one page to another casts a so-called vote, the weight of which depends on the weight of the pages that link to it. And we can't know their weight till we calculate it, so the process goes in circles. There should be hell of a formula to calculate it.

However, it's much simpler than it seems.

Calculation of PageRank

Math is so cool that you can use a simple iterative algorithm and calculate the PageRank score of a page without knowing the value of other pages that link to it. How does it work? Each time we run the calculation, we are getting a closer estimate of the final value. We remember each calculated value and repeat the calculations a number of times till the numbers stop changing much.

In order to prevent some pages from having too much influence, the PageRank formula also uses a damping factor. According to the theory, there's an imaginary surfer who is randomly clicking on links, gets bored at some point, and stops clicking. The probability that this person will continue clicking at any step is a damping factor. In the formula, the total value of pages is damped down by multiplying it by 0.85 (a generally assumed value).

It's also considered that the average sum of all pages equals one. Thus, even if a page has no backlinks (i.e., no votes), it still gets a small score of 0.15 (one minus a damping factor).

It's believed that Google recalculates PageRank scores after each crawl of the Web. As it expands, the initial approximation of PageRank decreases for all documents. Most probably, PageRank favors older pages as new pages cannot have a big profile of quality backlinks, so they receive lower scores.

If you want to dig deeper into the algorithm, refer to this article by Ian Rogers that remains one of the best on the subject. As for the calculation details, read a detailed and digestible explanation of the calculation process by Dixon Jones.

The history of public PageRank (infographics)

PageRank Infographics from SEO PowerSuite

As back as 2019, rumors had it that the PageRank algorithm hadn’t been used since 2006. In fact, the original PageRank patent had been replaced by another one relying on a set of seed sites. Those were sets of high-quality, reliable, and diverse pages, well-connected to the rest of the web with outgoing links. The ranking algorithm calculated the score of other web pages on the basis of their proximity to the seed sites, finding the shortest distance from seed sites to other pages via links.

The former Googler explained that the search engine needed a faster way to calculate the ranking scores, as the web rapidly grew from around 10 M pages to 150 B. So in fact, when we talk about the modern-day PageRank algorithm, we mean the same good old PageRank remastered.

Google finally retired the PageRank toolbar in 2016, hiding its scores from public access. Why? Well, because two can play at that game. PageRank gave rise to the link economy where purchased links were used to manipulate rankings, and at the start of the ranking system, it was dead simple.

The PageRank flow got its value. Passing authority from one page to another, which became informally known as passing link juice, was easily done through buying and selling links.

Advanced SEOs used black-hat techniques to manipulate rankings, commonly referred to as spamdexing. The manipulative arsenal varied from less vicious PageRank sculpting with the help of nofollow tags to heavy link farms and paid blog networks (PBNs). Google took the PageRank toolbar out of sight. Alongside, it worked on facilitating quality rather than quantity of links, nailing it down in its quality guidelines against link schemes.

Is the PageRank algorithm applied today?

PageRank stays Google’s secret sauce ingredient. When something is invisible, it doesn't mean it doesn't exist. The same goes for PageRank. It still matters for rankings, as it helps the search engine determine the most relevant result for a particular query.

Gary Illyes confirms in a tweet Google uses PageRank

What Google re-confirmed in 2020:

"Yes, we do use PageRank internally, among many, many other signals. It's not quite the same as the original paper, there are lots of quirks (eg, disavowed links, ignored links, etc.), and, again, we use a lot of other signals that can be much stronger."

John Mueller

What's more, PageRank hasn't gone invisible in the minds of SEOs. You know how it goes: give something to an SEO, and he or she will reverse-engineer it. The exact same thing happened to PageRank. SEOs just didn't want to stay blind to the important things happening behind the scenes. This way, a bunch of SEO houses (e.g., Moz, Ahrefs, Majestic, etc.) took the original PageRank formula as inspiration and developed alternatives to the metric.

We in SEO PowerSuite also gave it a try and developed our own alternative — InLink Rank. Of course, we can never say that PageRank and InLink Rank are the same metrics. The only thing we can say for sure — InLink Rank is based on the original Google PageRank formula. The catch is in the fact that we don't really know how exactly Google implements the formula.

We ran an experiment to find out whether Google still uses PageRank and which of the existing alternatives matches Google results best. As PageRank scores are generally associated with search positions, we've explored the possibility of a correlation between the alternative metric and SERPs. Let's look at the methodology and results.

InLink Rank and rankings correlation study

During three days, March 4-6, 2019, we looked at 30 Google search positions for 33,500 queries to get a list of 1,000,000 URLs. We recorded Domain InLink Rank for each URL. As a result, we had the raw data of one million rows to analyze.

Methodology of InLink Rank and PageRank score correlations

Further on, we compared this data with the correlations for Domain Authority by Moz, Citation Flow and Trust Flow by Majestic, and Domain Rank by Ahrefs taken from a study by Moz.

As a result, we've been able to draw two conclusions:

  1. There is a high correlation between Google search positions and Domain InLink Rank – 0.128.

Analyzing the raw data, we saw that URLs with lower rankings tended to have lower Domain InLink Rank, and vice versa. Of course, it does not mean it always works like that — correlation does not imply causation.

  1. SEO PowerSuite's InLink Rank has shown a 6% higher correlation compared to the next best metric — Domain Authority by Moz.

According to the similar analysis run by Moz, the strongest correlations between metrics and rankings were these (plus SEO PowerSuite's correlation):

A comparative study of ranking scores tweaking PageRank score calculation

While there's clearly a pleasant 6% stronger correlation with Domain InLink Rank comparing to the next best of Moz' Domain Authority, we should note that our experiment hasn't been performed using to the exact same methodology. It also doesn't mean InLink Rank is an absolute factor — the correlation can vary for different niches.

However, you can see that the secret sauce ingredient wasn't so secret after all — we were able to crack it and come close to Google's calculations. It means you can safely use Domain InLink Rank as a reliable metric for evaluating the ranking potential of a domain.

Note:

1. While Google PageRank, now invisible to us, is still at the core of Google's ranking mechanism, none of the industry metrics used in the analysis is directly used by Google and has any impact on rankings.

2. While PageRank alternatives used in the experiment are a good starting point for evaluating the ranking potential of domains, focusing on PageRank alone has never been a good SEO strategy. Likewise, a singular focus on any other metric won’t work out.

How to boost your PageRank score

Knowing that PageRank is still a part of the ranking mechanism is one thing. A completely different thing is to know how to influence it. Of course, no action can guarantee the increase of the actual PageRank score. However, we can monitor our SERP performance using alternative PR metrics (like InLink Rank) calculated according to the same approach.

Plus, we know two factors that hugely influence page's credibility (thus, boosting PageRank flow to it):

  • Quality backlinks
  • Efficient internal linking

Let's see how to optimize for each of these factors.

Quality backlinks

Backlinks carry link juice to your site. However, you need to make sure that this link juice is of high quality, otherwise, it can be harmful. To check links pointing to your site, run a regular backlink audit. Here, bear in mind that PageRank flows between pages, not domains. It means that links from a high-authority page on a low-authority site have more weight than links from a low-authority page on a high-authority site.

Check InLink Rank

To make sure your backlinks won't harm your site, check their InLink Rank scores in SEO SpyGlass. Run the tool, open or create a project for your site, and jump to the Backlinks tab.

Select those backlinks you'd like to check, click the Update button, and check the Backlink Authority box.

Check your pages' ranking potential

Once the analysis is complete, analyze the InLink Rank and Domain InLink Rank columns:

InLink Rank is a score in SEO PowerSuite replicating PageRank

When the value of any backlink is low (marked by a red dot), it can be a strong indicator of a spammy or low quality link. You should run an additional check (described below). Note that new sites and small local businesses tend to have links with lower PageRank values, so you should consider getting rid of such links only when all the factors point to the fact they are spammy.

Check for risks and errors

When you want to run an additional check for risks and errors, you need to pay attention to the page's authority, nofollow tag, and anchor text. All these factors can be checked in SEO SpyGlass.

1) Go to the Penalty Risk dashboard, select those links you're suspicious of, right-click them, and hit Update Penalty Risk.

Check the penalty risk score

In case you see some links having high Penalty Risk score, click on the “i” button to view Detected Penalty Risk factors:

Penalty risk factors

2) In the same tab, check the Links Back column for the dofollow attribute, as it might be switched off:

Check the dofollow and nofollow types of links

3) Check the Anchor Text column in the Backlinks tab to make sure the texts haven't been changed:

Anchor text describes what the linked page is about

An audit of HTTP responses of your pages is also a must-have in your backlink audit (if the column is not visible in your workspace, click anywhere in the columns' header to add it). This analysis will make you a plan of actions to safeguard your ranking scores through backlinks:

  • Fix 404 pages;
  • Revise 301 and 302 redirects;
  • Reach out to remove or disavow backlinks with a high penalty risk;
  • Whenever possible, ask the linking domains to optimize anchors.

Leverage linkless mentions

With a widespread use of a nofollow tag, how are search engines supposed to evaluate pages in terms of safety and relevance? Still carefully but we have a right to say that linkless brand mentions can be the future of link building. How come? Well, building linkless mentions is based on the same principles of credibility as building links. Plus, Google uses online brand mentions in its search algorithm. It evaluates your authority by looking at all the mentions of your brand and the reputation around them.

Luckily, growing brand mentions is much easier than building links. Here's what you need to do:

1) Find all mentions of your or your competitor's brand from all over the web by creating an alert for a brand in a social monitoring tool, like Awario:

Find linkless mentions and ask them to link

2) When you understand who is talking about your or your competitor's brand online, use the following tactics:

  • Jump into conversations and mention your brand whenever it's suitable (e.g., when your product or service can come in handy in a given situation);
  • Try to keep your customers engaged to make them talk about your brand (e.g., solve their problems, run contests, etc.);
  • Track industry keywords to understand your niche better and find those who can spread the word about your services to your audience.

Honestly, don’t be shy to use white-hat link-building tactics, they fall in line with Google’s quality guidelines and so far have proved to improve ranking scores.

Efficient internal linking

Link building is a time-consuming activity, and it can get out of hand. Thus, backing up your PageRank flow by efficient internal linking is a nice strategy, as internal linking is totally in your hands.

  • Keep your site structure shallow

As confirmed by John Mueller during one of the Google Webmaster Central hangouts, keeping any page 2-3 clicks away from the homepage is highly recommended. If your site is so huge that 2-3 clicks cause a nervous laughter, use breadcrumbs and internal search to make your site's browsing easier.

  • Make sure there are no orphan pages

Orphan pages are those not connected to any other page on your site and thus invisible to Google. It means no link juice for them. The most important thing is to check that no important pages of yours have become orphans.

You can either check the click depth of your pages in WebSite Auditor's Site Audit dashboard or go straight to the Visualization tab. Once your site structure is visualized, check for the nodes that are not connected with any others. These are your poor orphans.

Links form orphan pages don't pass any link juice

  • Add links to your page's main content

When you interlink content within your site, mind that link location influences its SEO value. Links placed within main content are considered to be more valuable than those in the navigation elements.

  • Include keywords in your anchor texts

Tell your readers what to expect if they follow your hyperlinked text.  The original PageRank paper states that “anchors often provide more accurate descriptions of web pages than the pages themselves”. Definitely, anchor text is associated with the page the link point to. Besides, it may help to rank resources uncrawlable by the text-based search bot, such as images or documents. So, one or two keywords in your anchor text won't be viewed as keyword stuffing. The same goes for alt attributes for your images that serve as anchors for text.

  • Keep an amount of links per each page reasonable

Too many links coming from a single page may look like selling links, and such a page may earn an algorithmic penalty or a manual action. Google Webmaster Guidelines say the number of links per page should be reasonably limited. It maintains the readability of pages and protects your pages against spam flagging. Plus, remember if you link to one and the same URL a few times on the same page, only the first anchor text will have an SEO value.

There is a widely-shared myth among SEOs that linking out to external resources hurts your site’s SEO because each outbound link sort of dilutes the PageRank on your site.

Avoiding the issue of PageRank, John Mueller remarks in a short video Q&A that “linking to other sites is a great way to provide value to your users”. Therefore, what is useful for users should also be useful for SEOs. However, he warns to watch out for outbound links that might look unnatural and be a sign of an artificial link:

  • Link exchange;
  • Advertisement;
  • Links in comments.

To avoid algorithmic or sometimes manual penalties, there are special tags such as ‘nofollow’, ‘sponsored’, and ‘UGC’ (user-generated content) to mark these links accordingly.

However, having natural links (i.e. links in content) to reputable sources that back your statements, let users investigate your sources, or learn more on the topic is deemed to be beneficial to your SEO.

Conclusion

PageRank definitely made a dent on the world as it helped Google become the search giant, and it still remains a part of its search algorithm. However, it's important to remember there are hundreds of other ranking factors. What makes pages rank higher is a combination of the factors, which can be different for each page. There's no sense in focusing on just one factor. Instead, take a holistic approach to your SEO activities to find a happy medium.


The article was originally published in April 2019 and has been revised to give up-to-date information. While Google rolls out its link spam update this fortnight, it is interesting to refresh our knowledge on PageRank and the value of links.