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 our Search index to present the
most relevant, useful results, all in a fraction of a second.
This page is a guide to understanding some of our 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.

Our ranking systems are designed to work on the page level, using a variety of signals and
systems to understand how to rank individual pages. Site-wide signals and classifiers are also
used and contribute to our understanding of pages. Having some good site-wide signals does not
mean that all content from a site will always rank highly, just as having some poor site-wide
signals does not mean all the content from a site will rank poorly.

We regularly improve our ranking 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.

You can also visit our How Search Works site
to understand how our ranking systems,
combined with other processes, work together so that Google Search delivers on our mission to
organize the world’s information and make it universally accessible and useful.

BERT

Bidirectional Encoder Representations from Transformers (BERT)
is an AI system Google uses that allows us 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: Our systems work to understand when people are seeking
  information about personal crisis situations to display hotlines and content from trusted
  organizations for certain queries related to suicide, sexual assault, poison ingestion,
  gender-based violence, or drug addiction. Learn more about how
  personal crisis information is displayed in Google Search.
• SOS Alerts: During times of natural disasters or wide-spread crisis
  situations, our 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. Learn more about
  how SOS Alerts work and how they’re part
  of Google’s crisis alerts
  that help in times of floods, wildfires, earthquakes, hurricanes, and other disasters.

Deduplication systems

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. Learn more about
how deduplication works and how to see omitted results
if desired, when deduplication happens.

Deduplication also happens with featured snippets.
If a web page listing is elevated to become a featured snippet, we 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

Our ranking systems consider the words in domain names as one of many factors to determine if
content is relevant to a search. However, our exact match domain system works to ensure we 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. Our
system adjusts for this.

Freshness systems

We have 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.

Link analysis systems and PageRank

We have 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 our core ranking systems used when Google first launched. Those curious can
learn more by reading the original PageRank research paper
and patent.
How PageRank works has evolved a lot since then, and it continues to be part of our core
ranking systems.

Local news systems

We have systems that work to identify and surface local sources of news whenever relevant,
such as through
our “Top stories” and “Local news” features.

MUM

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 we display.

Neural matching

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

We have 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 us 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 we process a significant
volume of such removals involving a particular site, we use that as a signal to improve our
results. In particular:

• Legal removals:
  When we receive a significant volume of
  valid copyright removal requests
  involving a given site,
  we are able to use that
  to demote other content from the site in our results. This way, if there is other
  infringing content, people are less likely to encounter it versus the original content. We
  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 significant
  proportion of CSAM content.
• Personal information removals:
  If we process a significant volume of personal information removals involving a site with
  exploitative removal practices,
  we demote other content from the site in our results.
  We also look to see
  if the same pattern of behavior is happening with other sites and, if so, apply demotions to
  content on those sites. We may apply similar demotion practices for sites that receive a significant
  volume of removals of content involving
  doxxing content,
  explicit
  personal imagery created or shared without consent, or
  explicit
  non-consensual fake content.

Passage ranking system

Passage ranking
is an AI system we use to identify individual sections or “passages” of a web page to
better understand how relevant a page is to a search.

RankBrain

RankBrain
is an AI system that helps us understand how words are related to concepts. It means we 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, our systems automatically display
content advisories
about rapidly-changing topics or when our 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. Learn more about our approach to delivering high-quality
information in Search.

Reviews system

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

Our site diversity system works so that we generally won’t show more than two web page listings
from the same site in our top results, so that no single site tends to dominate all the top
results. However, we may still show more than two listings in cases where our systems determine
it’s especially relevant to do so for a particular search. Site diversity generally treats
subdomains as part of a root domain. IE: 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 us from showing the most helpful and relevant results. We
employ a range of spam detection systems,
including SpamBrain, to deal with
content and behaviors that violate our spam policies.
These systems are constantly updated to keep up
with the latest ways that the spam threat evolves.

Retired systems

The following systems are noted for historical purposes. They’ve either been incorporated into
successor systems or made part of our core ranking systems.

Helpful content system

Announced in 2022 as the
“Helpful Content Update”, this was a system 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. In March 2024, it evolved and became part of
our core ranking systems, as our systems use a variety of signals and systems to
present helpful results to users.

Hummingbird

This was a major improvement to our overall ranking systems made in August 2013. Our
ranking systems have continued to evolve since then, just as they had been evolving before.

Panda system

This was a system designed to better ensure high-quality and original content was appearing
in our search results. Announced in 2011
and given the nickname of the “Panda,” it evolved and became part of our core ranking systems in 2015.

Penguin system

This was a system designed to combat link spam. Announced in 2012
and given the nickname of the “Penguin Update”, it was integrated
into our core ranking systems in 2016.