What is PageRank? How Google's Algorithm Began

Learn what PageRank is, how the original link-counting algorithm worked, and what actually survives of it inside Google's modern ranking systems.

PageRank is the link-analysis algorithm Google was founded on. It scores a page's importance by looking at who links to it, treating each link as a vote whose weight depends on the importance of the page casting it. Larry Page and Sergey Brin described it in their 1998 Stanford paper on the original Google prototype, and the name is a pun on both "web page" and Larry Page, which the paper era found funnier than we do.

The idea in one line, a page is important if important pages link to it. Circular on purpose, computed by iteration, and in 1998 it was the difference between Google and every keyword-matching engine before it.

How the Original Algorithm Worked

The classic formulation is the random surfer model. Imagine someone endlessly clicking links at random. On each page they either follow one of its outgoing links, or, with some probability, get bored and jump to a random page anywhere on the web. The paper set that damping behavior with a factor of 0.85, meaning an 85% chance of continuing to click. A page's PageRank is the fraction of time this eternal surfer spends on it.

Three properties fall out of the math, and they're the part worth remembering because they still shape how link value gets reasoned about:

Links from important pages are worth more. A vote from a heavily-linked page carries its accumulated weight.
A page's vote splits across its outgoing links. Link out to 200 pages and each one gets a sliver; link to 3 and each gets a real share. This is the origin of all the link equity intuitions SEO carries around.
Everything is relative. PageRank scores only mean anything compared against the rest of the graph.

The iteration runs until the numbers settle. Elegant, and famously gameable, which is roughly the history of the next decade of SEO as I understand it.

For years Google exposed a public 0 to 10 PageRank score in its browser toolbar. It became THE currency of SEO. Link marketplaces priced by it, sidebar widgets bragged about it, "PR5 site" was a product. Google stopped updating the public score in the mid-2010s and removed it from view in 2016, largely because the visible number was fueling the exact link economy the spam team was fighting.

The hangover is that the industry replaced it with third-party imitations, Domain Authority, Domain Rating and friends. Useful comparative shorthand, but vendor estimates rather than anything Google reads. Worth saying plainly since clients still ask for "our PageRank" and the honest answer is that nobody outside Google has seen a real one in a decade.

Does PageRank Still Exist?

Yes, in lineage if not in 1998 form. Google representatives have confirmed over the years that link-based importance signals descended from PageRank remain part of ranking, alongside hundreds of other signals and now a stack of machine-learned systems. The original algorithm also spawned refinements the public only glimpses through patents, and nofollow, introduced in 2005, exists specifically to let links opt out of passing this kind of value.

What's changed is weight and sophistication. Modern link evaluation cares about relevance, placement, and trust in ways raw PageRank never did, and content quality systems can override what links alone would say. Links still matter, any off-page SEO work rests on that, but the single-number era is over.

I find PageRank most useful now as a mental model rather than a metric. It explains why internal linking concentrates value, why a link from a strong page beats ten from weak ones, and why orphan pages starve.

What PageRank Thinking Gets You Today

Internal link architecture. Your site is a link graph too. Pages you link to often and from strong pages accumulate internal value, and pages nobody links to get nothing. This is the practical, fully-controllable descendant of the algorithm, covered in the internal linking entry.
Link equity budgeting. A page linking out to hundreds of targets dilutes each vote. Nobody's suggesting you count links like it's 2009, but a "related articles" module with 60 entries is doing less per entry than one with six.
Redirect hygiene. Redirects exist partly to preserve accumulated link value when URLs change. Letting old URLs 404 throws away votes the site already earned.

Sane link building. The random surfer doesn't care about your intent, only the graph. Links from pages that themselves have no incoming links pass approximately nothing, which is why fabricated networks of empty sites keep failing.

Common Mistakes to Avoid

Chasing third-party authority scores as if they were PageRank. They're estimates for comparison, not targets. Optimizing a vendor metric directly is cargo cult with dashboards.
PageRank sculpting with nofollow. The 2009-era trick of nofollowing internal links to steer value died when Google changed how nofollow's share was counted. Structure your links for readers.
Assuming all links vote equally. The entire point of the algorithm is that they don't, and every generation of link spam rediscovers this the expensive way.
Treating PageRank as dead because the toolbar is. The public number died. Link-based importance didn't, and behaving otherwise mostly produces sites with terrible internal linking.

In Practice

I think reading the original paper is genuinely worthwhile, it's short, readable, and slightly surreal, two grad students describing the prototype of the thing that reorganized the web, complete with worries about advertising-funded search engines being biased, in a paper that founded the biggest one. For day-to-day SEO, the model earns its keep every time you decide where a new page should be linked from. Ask where the eternal surfer already spends time on your site, and link from there.

What is Link Equity? - the modern working vocabulary for what PageRank measured.
What is Internal Linking? - the controllable link graph on your own site.
What is Domain Authority? - the third-party stand-in the toolbar era left behind.
What is Nofollow? - the attribute that lets links decline to vote.
What are Backlinks? - the raw material the algorithm counts.

Key Takeaways

PageRank scores importance by weighted links, votes from important pages count more, and each page's vote splits across its outgoing links.
The public toolbar score is long gone; third-party authority metrics are estimates, not the real thing.
Link-derived importance still exists inside Google's stack, refined and rebalanced among many other systems.
The model's real value today is guiding internal linking, redirects, and judging which external links can actually pass anything.