The Anatomy of a Large-Scale Hypertextual Web Search Engine

The short version: Links from pages that themselves have many links are worth more, and a page that links out to hundreds of others passes little to each.

What they studied

The paper that described the first version of Google: how it crawled and indexed the web, and how it ranked pages with PageRank and link text.

Key findings

• PageRank scores a page by the scores of the pages linking to it, with each linking page's score shared across its outgoing links.
• It models a "random surfer" who follows links and sometimes jumps to a random page, with a damping factor usually set to 0.85.
• Anchor text was attached to the page a link points to, because link text often describes a page better than the page describes itself.
• An appendix warned that search engines funded by advertising would be pulled toward advertisers rather than users.

Key learnings

• Links from pages that themselves have many links are worth more, and a page that links out to hundreds of others passes little to each.
• The words other sites use to link to you still describe you to search engines. Earn descriptive links.

Things to keep in mind

• Google's ranking today uses many more systems than PageRank. This paper explains the foundations, not the current algorithm.

Read it alongside

• 96.55% of pages get no traffic from Google (Ahrefs, 2023)
• 11.8 million Google results: what correlated with ranking (Backlinko, 2016)

Source

Academic paper (historic) by Sergey Brin and Lawrence Page, published by Stanford University, WWW7 conference, 1998.

• https://courses.cs.washington.edu/courses/cse454/15wi/papers/google.html

We read the source and checked every figure on this page against it on October 13, 2026. Summaries and learnings are ours. Spotted an error? Tell us.

More research like this in SEO Insights Central.