Rand Fishkin of SparkToro, a company that develops market research and audience analysis tools, received an email from a person claiming to have access to a massive leak of API documentation from Google’s search department. This was reported by 9to5Mac.

In this documentation, Google explains how the search system works, which components influence the generation of results, and much more. In total, the documentation is more than 2,500 pages long and contains 14,014 attributes (API functions) that come from Google’s internal “Content API Repository.”

This leak likely originated from GitHub, where the company appears to have accidentally made the documentation publicly available for the period from March 27 to May 7. While this information shows what data and factors can be used to generate results, their importance and priority is not disclosed.

Among the details of the document that Fishkin revealed is that the company decides which sites can be displayed at the top of the search for topics such as election or travel information.

The company has a kind of whitelist, for which sites are marked with isCovidLocalAuthority and isElectionAuthority. In particular, in 2019, during the outbreak of the Covid-19 pandemic, the company also chose which sites with information about the coronavirus to display first.

Fishkin also said that the documents show that Google calculates several types of metrics that can be called up by Chrome views associated with both individual pages and domains.

This means that the company tracks which pages browser users go to the most and offers them in the form of the most popular or most important pages on the site, although it was previously claimed that no data from Chrome is used.

In addition, it became known that the company uses the ratings of its platform to check the quality of EWOK to form a search. It is not yet clear what impact these ratings have on search results, but evidence of its use was also found in the documentation.

The last point that Fishkin shared is that Google also uses user click-through data to decide how much weight certain links have in the rankings. So, if certain pages on websites have no clicks, they fall into the low quality index and the link is ignored, while if pages have many clicks from verified devices, they have a higher quality index and become reliable.

Also, after SEOs began to familiarize themselves with the documentation, they claim that it contradicts Google’s public statements about how search works.

The company itself has not yet commented on the situation.