Patent-backed SEO
Deep dives, myth-busting, and practical patterns. Every article is anchored to a real Google patent you can read yourself.
30 articles
Navboost re-ranks Google results with click data. We read patent US 8,661,029 line by line: long clicks, short clicks, the last click, and what they mean for your pages.
Patent US 8,682,892, signed by Navneet Panda, scores a whole site by the ratio of independent links to searches for that site. Here's how the ratio works and what it changes for you.
Patent US 9,767,157 shows how Google predicts the quality of a site it has never scored, by comparing the phrases on its pages with sites it already rated. The full mechanism, step by step.
Patent US 7,536,408 describes how Google indexes documents by phrases, not words: good phrases, related phrases, clusters and information gain. The full mechanism and what it means for topical coverage.
Patent US 7,716,225 replaces PageRank's random surfer with a reasonable surfer who clicks some links more than others. The link features Google weighs, how the model learns from clicks, and what it changes for your linking.
Patent US 9,165,040 ranks pages by their link distance from a set of trusted seed sites. How seeds are chosen, how link length is computed, why the k-th nearest seed matters, and what it changes for your links.
Patent US 7,603,350 ranks search results by the trust placed in the people who labeled them. Labels, trust ranks, trust factors and topic-specific trust: the full mechanism and what it means for your authority.
Patent US 11,769,017 describes how Google writes an AI summary from search results, which documents it reads, how it links each sentence to a page that verifies it, and when it trusts the answer. The full mechanism and what it means for GEO.
Patent US 11,354,342 scores each document by the new information it brings beyond the pages a user has already seen, then re-ranks results as the user reads. The full mechanism and how to write content with real information gain.
Patent US 9,031,929, co-signed by Navneet Panda, scores a whole site by the ratio of searches that name it to searches that lead to a click on it. The formula, its variants, and what it means for your brand.
Patent US 9,959,315 adjusts the score of candidate answer passages with the path of headings above them, the match between headings and the question, the coverage of the passage, a preceding question and lists. The full mechanism and how to structure pages for featured answers.
Patent US 9,940,367 describes how Google builds candidate answer passages from the top results and scores them with query terms, expected answer terms and 6 query-independent signals. The full mechanism and how to write passages that get picked.
Patent US 9,213,748 describes how Google picks related questions for a query: topic sets from the queries that lead to each result, a database of past questions, equivalent questions merged through a graph, and answers that promote a question. The full mechanism and how to use it.
Patent US 8,577,893, signed by Anna Patterson and Paul Haahr, fingerprints the rare words around every link pointing to a page, then ranks the page by the number and distribution of those contexts. The full mechanism and what it means for link building.
Patent US 8,407,231 scores a page by how its links appear and disappear over time: a falling rate of new links marks it as stale, while sudden spikes can signal bought links. The full mechanism and what it means for link building.
Patent US 6,526,440, signed by Krishna Bharat, re-ranks the top results of a query (for example the first 1,000) by the links they give each other, after removing links from the same host, subnet or affiliated sites. The formula and what it means for topical links.
Patent US 8,046,371 scores local businesses inside a searched area by location prominence (mentions, reviews, directory listings, authoritative pages) instead of distance alone. The 5 factors, how they are combined and what they mean for local SEO.
Patent US 7,565,358 scores the agents (authors, editors, reviewers, publishers) who sign content, then uses their reputation to rank what they sign. How signatures, roles, topic scores and the agent graph work, and what it means for authorship today.
Patent US 8,150,842, written around Google Knol, scores authors from the reviews of their content, verifies their credentials, and uses both to rank what they write. The reputation factors, the credibility factor and what they mean for E-E-A-T.
Patent US 8,554,769 scores pages with a language model and a query stuffing score, then removes or demotes gibberish. How the 5-gram model, the query index and the thresholds work, and what they mean for scaled and AI content.
Patent US 8,938,463 builds a prior model of how often a result gets clicked because of how it is displayed, then compares actual clicks with that prediction. The bias features, the formula and what it means for click-based ranking.
Patent US 9,002,867 keeps the click statistics of every version of a page and weights each one by how much the content has changed since. How the difference is measured, when old clicks still count, and what it means before you rewrite a page.
Patent US 8,924,379 classifies pages as new or old from the queries that lead to them, tracks how query refinements change over time, and boosts pages during recurring popularity spikes. The 3 mechanisms and what they mean for fresh and seasonal content.
Patent US 9,792,330 identifies reviewers who are experts in a type of business in a given area, from the number of their reviews, then ranks local results with their ratings. How expertise scores, specificity levels and personalized experts work, and what it means for local SEO.
Patent US 7,636,714 learns synonyms from how users rewrite their own queries within a session, and validates each synonym in the context of the surrounding words. How pseudo-queries, shared results and context tests work, and what they mean for keyword targeting.
Patent US 9,477,759 answers questions like "Who is the king of Spain?" by collecting the entity references found in the top results, ranking them by frequency and topicality, and returning the winner. The full mechanism and what it means for being the source of a direct answer.
Patent US 7,716,216, signed by Georges Harik and Monika Henzinger, measures the distance between query terms in a page through its structure: implicit lists, headings and titles, not just word counts. The rules and what they mean for page structure.
“Older domains rank better” is one of SEO's most repeated myths. We trace it back to a misread patent and show what Google has actually said.
PageRank (US 6,285,999) ranks a page from the ranks of the pages that link to it. The exact formula, the worked example, the variants the patent describes for local links, and how to structure your internal links around it.
In 2012 Google introduced the Knowledge Graph as “things, not strings.” A Google patent shows how entity metrics and entity types can rank results. Here's what the shift to entities means for how you plan and write content.
The reference list of the Google patents behind our articles, from the oldest grant to the newest. Each line links to the patent on Google Patents and to our full analysis.
Want it done for you?
A free audit grounded in the same patents we write about.