Google Leaks - The 14 algorithm secrets we've learned

[shukla7789]

Rand Fishkin, former SEO and founder of the Moz portal, made a leak of documents allegedly from Google go viral. A few days later, an official response was received, confirming the authenticity of said documents and suggesting caution in not having all the context: "We recommend being careful with inaccurate assumptions about Search based on decontextualized, outdated or incomplete information."

The documentation, spread across more than 2,500 documents, has some records updated to 2023, so we could conclude that some of the parts are lebanon number dataset up to date. This information describes the data that Google's Search API works with, providing information on nearly 14,000 attributes and functions of Google's algorithm used to improve searches.

It is important to clarify that any conclusion is pure speculation and that we must be cautious with the information extracted from the documentation. As we always say, the best approach we can take is to test, experiment and analyze results to understand which is the best SEO course for our project.

Still, by looking at what Google considers important enough to store on its systems, we can come up with some interesting conclusions.

Relevant findings
After closely analyzing these documents, some interesting findings have been made.

Importance of clicks: Although we already knew this from the antitrust trial, it is now clear that Google uses user click data to rank websites, filtering out unwanted clicks and measuring click duration (bounce back to a new search). They use geotargeting and data from the Chrome browser to improve the accuracy of results.
NavBoost System: This highly important system for SEO uses the number of searches for a keyword to identify trends, the number of clicks on a result, and the ratio of long to short clicks. It also helps with alignment with search intent, for example by associating certain queries with a preference for images or videos. NavBoost is also used at the domain level to assess the quality of websites; the data and scores generated by this system are directly integrated into Google's search systems, likely aligned with the EEA.
Pre- and post-search analysis: Google analyzes and relates searches that follow the main search. For example, if I search for "SEO" and then perform a second search for "Smartup" and click on smartupdigital.com, it is possible that, if that pattern is detected, the Smartup page will start to rank for the "SEO" search, since NavBoost will understand that there is an association.
Chrome Browser: Uses cookie history and user browsing and click data to combat click spam (fraudulent clicks).
Link Ranking: Links are ranked as low, medium, and high quality based on click data, affecting PageRank transmission and spam detection.
Site Authority: Although Google avoids the term “domain authority,” it does measure site authority using a metric called “siteAuthority.”
Link value: Stratifies its index into levels, placing greater value on links from well-positioned pages or news portals.
EEAT and content: Evaluates the originality of content with metrics such as OriginalContentScore and titlematchScore. Author visibility is given importance to differentiate human content from automated content. The author is revealed to be relevant to the algorithm.
Video Focus: Sites with more than 50% of their pages containing video may receive special treatment in Google’s algorithms.
Small sites: Google labels “small personal sites” and treats them differently in search results, possibly to reduce their visibility compared to larger ones.
Penalties: There are penalties for domains that exactly match non-branded search queries, for example: best-luxury-watches.com.