A guide to Google Search ranking systems
Google uses automated ranking systems that look at many factors and signals about hundreds of billions of web pages and other content in their Search index to present the most relevant, useful results, all in a fraction of a second.
They regularly improve these systems through rigorous testing and evaluation and provide notice of updates to our ranking systems when those might be useful to content creators and others.
This page is a guide to understanding some of their more notable ranking systems. It covers some systems that are part of our core ranking systems, which are the underlying technologies that produce search results in response to queries. It also covers some systems involved with specific ranking needs.
Bidirectional Encoder Representations from Transformers (BERT) is an AI system Google uses that allows them to understand how combinations of words express different meanings and intent.
Crisis information systems
Google has developed systems to provide helpful and timely information during times of crisis, whether those involve personal crisis situations, natural disasters, or other wide-spread crisis situations:
- Personal crisis: Google systems work to understand when people are seeking information about personal crisis situations to display hotlines and content from trusted organisations for certain queries related to suicide, sexual assault, poison ingestion, gender-based violence, or drug addiction.
- SOS Alerts: During times of natural disasters or wide-spread crisis situations, the Google SOS Alerts system works to show updates from local, national, or international authorities. These updates may include emergency phone numbers and websites, maps, translations of useful phrases, donation opportunities, and more.
Searches on Google may find thousands or even millions of matching web pages. Some of these may be very similar to each other. In such cases, our systems show only the most relevant results to avoid unhelpful duplication.
Deduplication also happens with featured snippets. If a web page listing is elevated to become a featured snippet, they don't repeat the listing later on the first page of results. This declutters the results and helps people locate relevant information more easily.
Exact match domain system
Google ranking systems consider the words in domain names as one of many factors to determine if content is relevant to a search. However, the exact match domain system works to ensure they don't give too much credit for content hosted under domains designed to exactly match particular queries. For example, someone might create a domain name containing the words "best-places-to-eat-lunch" in hopes all those words in the domain name would propel content high in the rankings. This system adjusts for this.
Google has various "query deserves freshness" systems designed to show fresher content for queries where it would be expected. For example, if someone is searching about a movie that's just been released, they probably want recent reviews rather than older articles from when production began. For another example, ordinarily a search for "earthquake" might bring back material about preparation and resources. However, if an earthquake happened recently, then news articles and fresher content might appear.
Helpful content system
The helpful content system is designed to better ensure people see original, helpful content written by people, for people, in search results, rather than content made primarily to gain search engine traffic.
Link analysis systems and PageRank
Google has various systems that understand how pages link to each other as a way to determine what pages are about and which might be most helpful in response to a query. Among these is PageRank, one of their core ranking systems used when Google first launched. How PageRank works has evolved a lot, and it continues to be part of their core ranking systems.
Local news systems
Google has systems that work to identify and surface local sources of news whenever relevant, such as through their "Top stories" and "Local news" features.
Multitask Unified Model (MUM) is an AI system capable of both understanding and generating language. It's not currently used for general ranking in Search but rather for some specific applications such as to improve searches for COVID-19 vaccine information and to improve featured snippet callouts they display.
Neural matching is an AI system that Google uses to understand representations of concepts in queries and pages and match them to one another.
Original content systems
Google has systems to help ensure we are showing original content prominently in search results, including original reporting, ahead of those who merely cite it. This includes support of a special canonical markup creators can use to help Google better understand what is the primary page if a page has been duplicated in several places.
Removal-based demotion systems
Google has policies that allow the removal of certain types of content. If they process a high volume of such removals involving a particular site, they use that as a signal to improve the results. In particular:-
- Legal removals: When Google receives a high volume of valid copyright removal requests involving a given site, they are able to use that to demote other content from the site in the results. This way, if there is other infringing content, people are less likely to encounter it versus the original content. Google will apply similar demotion signals to complaints involving defamation, counterfeit goods, and court-ordered removals. In the case of child sexual abuse material (CSAM), we always remove such content when it is identified and we demote all content from sites with a high proportion of CSAM content.
- Personal information removals: If we process a high volume of personal information removals involving a site with exploitative removal practices, they demote other content from the site in the search results. They also look to see if the same pattern of behaviour is happening with other sites and, if so, apply demotions to content on those sites. Google may apply similar demotion practices for sites that receive a high volume of doxxing content removals. Furthermore, Google have automatic protections designed to prevent non-consensual explicit personal images from ranking highly in response to queries involving names.
Passage ranking system
Passage ranking is an AI system Google uses to identify individual sections or "passages" of a web page to better understand how relevant a page is to a search.
RankBrain is an AI system that helps them understand how words are related to concepts. It means Google can better return relevant content even if it doesn't contain all the exact words used in a search, by understanding the content is related to other words and concepts.
Reliable information systems
Multiple systems work in various ways to show the most reliable information possible, such as to help surface more authoritative pages and demote low-quality content and to elevate quality journalism. In cases where reliable information might be lacking, Googles systems automatically display content advisories about rapidly-changing topics or when their systems don't have high confidence in the overall quality of the results available for the search. These provide tips on how to search in ways that might lead to more helpful results.
The reviews system aims to better reward high quality reviews, content that provides insightful analysis and original research, and is written by experts or enthusiasts who know the topic well.
Site diversity system
The Google site diversity system works so that they generally won't show more than two web page listings from the same site in the top results, so that no single site tends to dominate all the top results. However, Google may still show more than two listings in cases where their systems determine it's especially relevant to do so for a particular search. Site diversity generally treats subdomains as part of a root domain (i.e. listings from a subdomain (subdomain.example.com) and the root domain (example.com) will all be considered from the same single site)). However, sometimes subdomains are treated as separate sites for diversity purposes when deemed relevant to do so.
Spam detection systems
No one wants their email inbox filled with spam, which is why spam filters are so helpful. Search faces a similar challenge, because the internet includes huge amounts of spam that, if not dealt with, would prevent Google from showing the most helpful and relevant results. They employ a range of spam detection systems, including SpamBrain, to deal with content and behaviours that violate our spam policies. These systems are constantly updated to keep up with the latest ways that the spam threat evolves.
The systems below are noted for historical purposes. They've either been incorporated into successor systems or made part of Googles core ranking systems.
This was a major improvement to the overall ranking systems made in August 2013. Googles ranking systems have continued to evolve since then, just as they had been evolving before.
This was a system designed to better ensure high-quality and original content was appearing in search results. Announced in 2011 and given the nickname of the "Panda," it evolved and became part of the core ranking systems in 2015.
This was a system designed to combat link spam. Announced in 2012 and given the nickname of the "Penguin Update", it was integrated into the core ranking systems in 2016.