Key takeaways
Google ranks pages using many interconnected systems, not a single algorithm. PageRank and link analysis weigh links, Hilltop adds topical authority, Hummingbird, RankBrain, BERT and MUM interpret query meaning, the Helpful Content signals (now folded into core ranking) reward people-first content, and Penguin and SpamBrain neutralize spam. They run together on every query.
- Google Search is a stack of ranking systems working together, not one monolithic algorithm.
- Link-based systems (PageRank, Hilltop) judge authority; language systems (Hummingbird, RankBrain, BERT, MUM) judge meaning.
- The Helpful Content System became part of Google's core ranking systems with the March 2024 core update.
- Penguin (link spam) and SpamBrain (Google's AI-based spam-prevention system) defend results from manipulation.
- Each system below links to its own dedicated deep-dive guide.
Why Google Search is a stack, not a single algorithm
In short
Google Search is not one algorithm but a stack of interconnected ranking systems that run together on every query. Link-analysis systems estimate authority, language systems interpret meaning, helpfulness signals judge whether a page serves people, and anti-spam systems strip out manipulation.
This hub maps the full stack and links to a dedicated deep-dive for each system, so you can branch off to whichever one you need.
People often talk about "the Google algorithm" as if it were one formula. In reality, Google Search runs on many ranking systems that operate together on every query. Google itself publishes a guide to these systems and distinguishes between long-running ranking systems (like PageRank, RankBrain and BERT) and one-time ranking updates (like a specific core update).
This pillar maps the full stack in the order it grew: first the link-analysis systems that estimate authority, then the language-understanding systems that interpret meaning, then the quality and anti-spam systems that keep results trustworthy. Each section below gives a one-paragraph summary and points to a dedicated spoke guide. Use this page as the canonical entry point and branch off to whichever system you need. If you want a deeper foundation on the link half of the stack, start with our companion guide to Google PageRank.
The ranking stack at a glance
- What it is
- Many ranking systems, not one algorithm
- Authority systems
- PageRank, link analysis, Hilltop
- Language systems
- Hummingbird, RankBrain, BERT, MUM
- Quality signals
- Helpful Content, now in core ranking
- Anti-spam systems
- Penguin, SpamBrain
- How they run
- Together, as layers, on every query
- Earliest system
- PageRank, Stanford, 1996
- Latest shift
- Helpfulness folded into core, March 2024
PageRank and link analysis: authority through links
PageRank is the link-analysis system that launched Google. Larry Page and Sergey Brin developed it at Stanford University in 1996, and it was described in their 1998 paper The Anatomy of a Large-Scale Hypertextual Web Search Engine. It models an imaginary random surfer clicking links, treating each link as a vote and weighting votes by the linking page's own importance, with a damping factor (commonly 0.85). The underlying patent, US 6,285,999, was assigned to Stanford, not Google.
Google still lists link analysis and PageRank as "one of our core ranking systems," while stressing it is just one signal among many. Read the dedicated guide: The Complete Guide to Google PageRank.
Hilltop: topical authority from expert pages
Hilltop refined the idea of authority by adding topical context. Created by Krishna Bharat and George A. Mihaila, it distinguishes expert pages (pages on a topic that link out to many non-affiliated relevant resources) from authority pages (pages that receive links from multiple expert pages). Google acquired the algorithm in February 2003.
Where raw PageRank counts links broadly, Hilltop asks whether the links come from credible, on-topic sources - an early step toward today's emphasis on relevance and authority together. Read the dedicated guide: The Complete Guide to Google's Hilltop Algorithm.
Hummingbird: the shift to meaning
Hummingbird, announced on September 26, 2013, was a wholesale overhaul of Google's core algorithm. Instead of matching individual keywords, it considers the context of words together so that pages matching the meaning of a query rank better. It marked Google's pivot toward semantic and conversational search and set the stage for the machine-learning systems that followed.
Hummingbird is the framework inside which RankBrain, BERT and other language systems later operated. The dedicated guide to Hummingbird is part of this same cluster of ranking-system explainers.
RankBrain: machine learning for unfamiliar queries
Google confirmed RankBrain on October 26, 2015. It is a machine-learning system that helps interpret queries, especially the many searches Google has never seen before, by mapping unfamiliar words and phrases to similar known concepts. At launch Google described it as one of the three most important ranking signals, alongside links and content.
RankBrain works inside Hummingbird's semantic framework and pushed SEO toward satisfying genuine search intent rather than exact keywords. Read the dedicated guide: The Complete Guide to Google RankBrain.
BERT and MUM: deep language understanding
BERT (Bidirectional Encoder Representations from Transformers) came from a 2018 Google AI paper and rolled into Search on October 25, 2019, initially affecting about 1 in 10 English-language queries in the US. By reading words in both directions of context, BERT understands how small words like prepositions and negations change meaning - for example correctly interpreting "2019 brazil traveler to usa need a visa."
MUM (Multitask Unified Model), announced on May 18, 2021, is described by Google as 1,000 times more powerful than BERT, multimodal (text and images), and trained across 75 languages. It powers select features rather than ranking every query. Read the dedicated guide: The Complete Guide to Google BERT and MUM.
The Helpful Content System: rewarding people-first content
Announced in August 2022, the Helpful Content System introduced an automated, site-wide signal designed to reward content created for people first and to down-rank content created mainly to rank in search engines. With the March 5, 2024 core update, Google integrated these helpfulness signals into its core ranking systems rather than running them as a separate system, and reported reducing low-quality, unoriginal content by roughly 45 percent.
Google's self-assessment questions (originality, depth, expertise, first-hand experience) map closely to its E-E-A-T guidance. Read the dedicated guide: The Complete Guide to Google's Helpful Content System.
Penguin and SpamBrain: defending the results
Penguin launched on April 24, 2012 to target link spam and manipulative backlink schemes. With Penguin 4.0 on September 23, 2016 it became real-time and part of the core algorithm; rather than penalizing whole sites, it now discounts spammy links so they pass no value.
SpamBrain is, in Google's own words, "our AI-based spam-prevention system." Google improves it over time to catch new spam types and uses it in link-spam updates to neutralize the effect of unnatural links. Together these systems keep authority and language signals from being gamed. Read the dedicated guide: The Complete Guide to Google Penguin and SpamBrain.
How the systems work together
On a single query these systems run as layers, not in isolation. Language systems (Hummingbird, RankBrain, BERT, MUM, neural matching) interpret what the searcher means. Authority systems (PageRank, link analysis, Hilltop-style topical signals) estimate which pages are credible. Quality and helpfulness signals - now part of core ranking - judge whether a page genuinely serves people. Anti-spam systems (Penguin, SpamBrain) strip out manipulation before results are shown.
- Meaning first: the query is understood semantically before candidates are scored.
- Authority and relevance together: links matter, but on-topic, trustworthy links matter more.
- Helpfulness as a core filter: people-first content is rewarded across the site, not page by page.
- Spam neutralized, not just penalized: manipulative links are discounted so they earn nothing.
That is why no single "hack" wins: improving one layer rarely helps if another is failing. The spoke guides above explain how to align with each. If you are weighing where to focus, the link-authority half is covered in depth in our RankBrain guide for meaning and the Penguin and SpamBrain guide for keeping links clean.
History of Google's ranking systems: a timeline
Google's ranking stack grew over nearly three decades, from a 1996 Stanford link-analysis project to the 2024 integration of helpfulness signals into core ranking.
-
1996
PageRank developed at Stanford
Larry Page and Sergey Brin build PageRank as a Stanford research project; the 1998 paper describes the random-surfer link model.
-
2003
Google acquires Hilltop
Google acquires the Hilltop algorithm (Bharat and Mihaila), bringing expert-and-authority-page topical signals into its systems.
-
2012
Penguin launches
On April 24, 2012 Penguin begins targeting link spam and manipulative backlink schemes.
-
2013
Hummingbird overhaul
Announced September 26, 2013, Hummingbird rebuilds the core algorithm around meaning and conversational queries.
-
2015
RankBrain confirmed
Google confirms RankBrain on October 26, 2015 as a machine-learning system for interpreting unfamiliar queries.
-
2016
Penguin 4.0 goes real-time
On September 23, 2016 Penguin becomes part of the core algorithm and discounts spammy links rather than penalizing whole sites.
-
2019
BERT in Search
On October 25, 2019 BERT rolls into Search, affecting about 1 in 10 English-language US queries at launch.
-
2021
MUM announced
On May 18, 2021 Google unveils MUM - multimodal, multilingual, and described as 1,000x more powerful than BERT.
-
2022
Helpful Content System
In August 2022 Google launches an automated, site-wide signal rewarding people-first content.
-
2024
Helpfulness folded into core
The March 5, 2024 core update integrates helpful-content signals into core ranking systems.
The major ranking signals at a glance
Across the whole stack, a handful of signal families do most of the work. Seeing them side by side shows why no single tactic wins on its own.
| Signal | What it captures |
|---|---|
| Link authority (PageRank) | Quality and quantity of links pointing to a page, weighted by the linking pages' own importance. Still a core ranking system per Google. |
| Topical authority (Hilltop-style) | Whether links and citations come from credible, on-topic "expert" sources rather than unrelated pages. |
| Query meaning and intent | Semantic interpretation by Hummingbird, RankBrain, BERT and MUM - including context from prepositions, negations and word order. |
| Helpfulness and people-first content | Originality, depth, expertise and first-hand experience; now part of Google's core ranking systems after March 2024. |
| Spam and link manipulation | Penguin discounts manipulative links; SpamBrain, Google's AI-based spam-prevention system, detects and neutralizes spam. |
How to align with the whole stack
Because these systems run as layers, the durable strategy is to satisfy each one at once: earn on-topic links, write for genuine intent, prove helpfulness and experience, and keep your link profile clean.
-
Earn links from on-topic, credible sites rather than chasing volume.
PageRank weighs link quality and authority, and topical relevance (the Hilltop idea) makes on-subject links count for more.
-
Write in natural language that answers the full intent behind a query.
Hummingbird, RankBrain and BERT score meaning, not keyword matches, so content that satisfies intent ranks better.
-
Run Google's helpful-content self-assessment on your pages.
Helpfulness signals are now part of core ranking, so original, expert, people-first content is rewarded site-wide.
-
Audit your backlink profile and disavow only genuinely manipulative links.
Penguin and SpamBrain discount spammy links automatically; clean, earned links are what carry ranking value.
-
Demonstrate first-hand experience and clear authorship.
Google's quality guidance treats experience and expertise (E-E-A-T) as central to high-quality, trustworthy content.
Ranking algorithm myths vs. reality
Few topics in SEO carry as much folklore as "the Google algorithm." Here are the most common myths and what is actually true.
Myth Google uses one master algorithm that you can crack.
Reality Search runs on many interacting systems - link analysis, language models, helpfulness and anti-spam - that Google updates continuously.
Myth E-E-A-T is a direct ranking factor you can score.
Reality Google states E-E-A-T is not itself a ranking factor; it is a framework describing the qualities of content that its systems aim to reward.
Myth RankBrain replaced the rest of the algorithm.
Reality RankBrain is one machine-learning system among many and works inside Hummingbird's framework alongside PageRank, BERT and others.
Myth Penguin penalizes your whole site for bad links.
Reality Since Penguin 4.0 in 2016, Google discounts the spammy links so they pass no value, rather than demoting the entire site.
Myth The Helpful Content System is still a separate system to optimize for.
Reality With the March 2024 core update, its helpfulness signals were folded into Google's core ranking systems.
Frequently asked questions
It is many. Google Search runs on a stack of interconnected ranking systems - including PageRank, RankBrain, BERT, MUM, helpful-content signals and spam detection - that operate together on every query. Google publicly documents these as separate ranking systems rather than a single algorithm.
PageRank is the link-analysis system Larry Page and Sergey Brin built at Stanford in 1996. It estimates a page's importance from the quality and quantity of links to it. Google still lists link analysis and PageRank as one of its core ranking systems, though just one of many signals.
All three help Google understand language. RankBrain (2015) interprets unfamiliar queries using machine learning. BERT (2019) reads word context in both directions to grasp nuance. MUM (2021) is multimodal and multilingual, described as 1,000 times more powerful than BERT, and powers select features.
Launched in August 2022, it was an automated, site-wide signal rewarding people-first content. With the March 2024 core update, Google integrated these helpfulness signals into its core ranking systems, so they are no longer a separate system but part of how core ranking works.
Penguin, launched in 2012, targets link spam; since 2016 it discounts manipulative links rather than penalizing whole sites. SpamBrain is Google's AI-based spam-prevention system, which it improves over time to detect new spam types and neutralize unnatural links.
No. Google states that E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) is not itself a ranking factor. It is a framework in Google's quality guidance describing the qualities of helpful, trustworthy content that its automated ranking systems aim to reward.
Language systems interpret what you mean, authority systems estimate which pages are credible, helpfulness signals judge whether a page genuinely serves people, and anti-spam systems strip out manipulation. Because they run as layers, no single optimization wins if another layer is failing.
Each section of this pillar links to a dedicated guide - covering PageRank, Hilltop, Hummingbird, RankBrain, BERT and MUM, the Helpful Content System, and Penguin and SpamBrain. Start with whichever system is most relevant to your current SEO question and branch out from there.
The bottom line
Bottom line
There is no single Google algorithm to crack. Search is a stack of systems that grew over 25 years: link analysis estimates authority, language models interpret meaning, helpfulness signals reward people-first content, and anti-spam systems strip out manipulation - all running together on every query. The durable strategy is to satisfy each layer at once, then deepen your knowledge of the one that matters most for your situation using the guides below.
References
- PageRank - Wikipedia
- The Anatomy of a Large-Scale Hypertextual Web Search Engine (Brin and Page, 1998)
- Hilltop algorithm - Wikipedia
- Google Hummingbird - Wikipedia
- RankBrain - Wikipedia
- Understanding searches better than ever before (BERT in Search) - Google
- MUM: A new AI milestone for understanding information - Google
- Creating helpful, reliable, people-first content - Google Search Central
- Google March 2024 core update and new spam policies - Google
- Google Penguin - Wikipedia
- Google Search spam updates and your site (SpamBrain) - Google Search Central
- A guide to Google Search ranking systems - Google Search Central