In brief
A leak of internal Google API documents, published on GitHub on March 13 and analyzed by Rand Fishkin and Michael King, describes 2,596 modules with 14,014 attributes used in ranking. The documents confirm that links and PageRank still count, that click metrics such as goodClicks and lastLongestClicks are tracked, that Chrome browser data is used, and that brand authority, freshness and title match with the query all matter.
- The documents list ranking features but not their weights, and the information is accurate as of March
- Google keeps a copy of every version of every indexed page but uses only the last 20 changes to analyze links
- Shorter content receives an originality score from 0 to 512, and a titlematchScore measures how well a page title matches the query
A massive leak of internal Google documents has exposed the search engine’s most intimate secrets, revealing some of the most important elements Google uses to rank content. This leak, which could be one of the biggest in SEO history, could fundamentally change the way industry professionals optimize their content.
What Happened
On March 13, thousands of documents from Google’s internal API were published to GitHub by an automated bot called yoshi-code-bot. These documents were shared with SparkToro co-founder Rand Fishkin, who, along with iPullRank CEO Michael King, analyzed and commented on the material. Here’s what we found.
Why This Leak Matters
We have gained unprecedented insight into how Google’s ranking algorithm might work, which is invaluable for SEO professionals. This leak provides us with crucial details about how Google ranks content in search results, providing a unique opportunity to adjust optimization strategies.
What the Documents Contain
Thanks to Fishkin and King’s analysis, here’s what we know about the internal documents:
- Current Status: Information is accurate as of March.
- Ranking Features: The documentation represents 2,596 modules with 14,014 attributes.
- Feature Weights: The docs do not specify how the different ranking features are weighted, just that they exist.
- “Twiddlers”: Re-ranking functions that can modify the information retrieval score or change the ranking of a document.
- Demotions: Content can be downgraded for a variety of reasons, including:
- A link does not match the target site.
- SERP signals indicate user dissatisfaction.
- Product reviews.
- Geographic location.
- Exact matching domains.
- Pornographic content.
- Change History: Google keeps a copy of every version of every indexed page, using only the last 20 changes to analyze links.
Key Ranking Factors
- Links Matter: Link diversity and relevance remain key. PageRank is still alive and considered for every document.
- Successful Clicks: Google uses various click metrics, such as badClicks, goodClicks, lastLongestClicks, and unsquashedClicks.
- Short Content Rated Highly: Shorter content gets scores based on originality, with a score from 0 to 512.
- Chrome data: A module called ChromeInTotal indicates that Google uses data from the Chrome browser for ranking.
- Brand Authority: Building a strong brand is more important than ever to improve organic rankings.
What Does It All Mean?
According to King, “you need to generate more successful clicks using a broader set of queries and get a greater diversity of links to continue to rank well. Very strong content will do this naturally. Focusing on generating qualified traffic and providing a better user experience will send signals to Google that your page deserves to rank.”
Confirmations from the Antitrust Trial
Some of the practices revealed in the documents had already come to light during the antitrust trial in the United States, where figures such as Satya Nadella of Microsoft testified against Google. Among these practices, the amount of clicks a page gets has been confirmed as a significant ranking factor, especially with Google’s Navboost system, which is one of the important signals used for ranking.
SEO Implications
This revelation will profoundly change SEO strategies. Professionals will have to:
- Create High-Quality, Short Content: Maximize the originality and impact of shorter content
- Build a Strong Brand: Invest in creating a recognizable and trustworthy brand.
- Optimize for Clicks: Improve user experience to increase successful clicks.
- Analyze Chrome Data: Understand how your browsing data affects your rankings.
Other Interesting Elements
According to internal Google documents:
- Freshness Matters: Google considers dates in the byline (bylineDate), in the URL (syntacticDate), and in the page content (semanticDate).
- Thematic Relevance: Google vectorizes pages and sites, comparing the page embeddings (siteRadius) with those of the site (siteFocusScore).
- Domain Registration Information: Google stores your domain registration information (RegistrationInfo).
- Page Titles Matter: Google has a feature called titlematchScore that measures how well a page title matches a query.
- Average Term Weight: Google measures the average weight of terms in documents (avgTermWeight) and anchor text.
Conclusions
The revelation of Google’s search engine secrets marks a turning point for the world of SEO. One thing is certain: the online search landscape will never be the same again.
Sources: https://searchengineland.com/google-search-document-leak-ranking-442617
Frequently asked questions
What did the Google search algorithm leak reveal?
The leaked internal API documentation describes 2,596 modules with 14,014 attributes that Google can use to rank content, without saying how each feature is weighted. It shows that links and PageRank still matter, that Google tracks click metrics such as badClicks, goodClicks, lastLongestClicks and unsquashedClicks, that Chrome browser data is used through a module called ChromeInTotal, and that re-ranking functions called twiddlers can modify a document's score or position.
How did the Google internal documents leak happen?
On March 13, thousands of documents from Google's internal API were published on GitHub by an automated bot called yoshi-code-bot. The material was shared with Rand Fishkin, co-founder of SparkToro, who analyzed and commented on it together with Michael King, CEO of iPullRank. Their analysis is the basis of what is known about the content, which is accurate as of March.
Does Google use clicks as a ranking factor?
Yes. The leaked documents list several click metrics, including badClicks, goodClicks, lastLongestClicks and unsquashedClicks, and the United States antitrust trial had already confirmed that the number of clicks a page receives is a significant ranking signal, in particular through Google's Navboost system. According to Michael King, pages need to generate more successful clicks across a broader set of queries, which strong content and a good user experience produce naturally.
What can get a page demoted in Google search according to the leaked documents?
The documents describe demotions for several reasons: a link that does not match the target site, SERP signals that indicate user dissatisfaction, product reviews, geographic location, exact match domains and pornographic content. The documents also describe twiddlers, re-ranking functions that can change the information retrieval score or the position of a document, so avoiding these conditions matters even when other signals are strong.
How should SEO strategy change after the Google algorithm leak?
Four adjustments follow from the documents: create high-quality short content, since shorter content is scored on originality; build a recognizable and trustworthy brand, because brand authority weighs more than ever on organic rankings; improve user experience to increase successful clicks; and understand how Chrome browsing data affects rankings. Freshness also counts, with Google reading dates in the byline, in the URL and in the page content.
Sources
- Google Search document leak reveals inner workings of ranking algorithmSearch Engine Land
Figures and dates come from the sources listed above. Last substantive revision: April 11, 2025.
