Article 15 min read
Google's Algorithm: The Systems It Lists, the Updates It Dates, and What No Outsider Knows
by Daniel Sócrates published
On September 14, 2026, we searched “algoritmo do google” (Google algorithm) on Google in Brazil, where we are based. At the top, the AI Overview answered in one sentence. The algorithm, it said, is a complex system with more than 200 factors. The source next to that sentence was a HubSpot blog post. 1
Right below came a page from Google itself. It talks about Search algorithms, in the plural, and gives no number of factors. 2 I read Google’s current pages on the subject. The number 200 appears on none of them. 2 3 4 5
What those pages do contain is something else. A list of named systems, a log of dated updates, and a page on how each change gets tested. This piece opens all three documents. Then it opens the 2024 leak, to separate what it showed from what no outsider knows.
What Google says decides the order among pages already has its own piece, on how to rank first on Google. Here the question comes one step earlier. Who, inside Google, puts that order together?
The hospital has departments
Think of a large hospital. Someone says the hospital decided. But who decided? The intake nurse flagged the emergency. The lab read the blood work. Radiology found the fracture. The pharmacy checked the dose.
Each department has a name and a job of its own. And the hospital changes all the time. Sometimes it tweaks a small protocol, and nobody notices. Sometimes it closes a wing for renovation, with a start date and an end date. Before changing a major protocol, it tries it out on a single floor.
Google Search is organized in a similar way. “The algorithm” is a nickname for a set of programs, each with a job. Google calls each of these programs a system. This is called a ranking system.
The official list: seventeen named systems
Google keeps a page called “A guide to Google Search ranking systems.” It was last updated on December 10, 2025. 3
The first sentence sets the scale. The systems look at many factors and signals across hundreds of billions of pages, in a fraction of a second. 3 The second sentence sets the limit. The guide covers some of the more notable systems. 3
Notice the word “some.” The list is partial, and Google itself says so. Some of the systems belong to the core, the technologies that produce results for any search. Others serve specific ranking needs. 3
On September 14, 2026, the active list had seventeen entries. 3 I grouped the seventeen by job, and the grouping is mine.
| Job | Systems, by the guide’s names | What the guide says |
|---|---|---|
| Understand language | BERT; MUM; neural matching; passage ranking; RankBrain | Understand meaning and intent, match concepts in the search to those on the page, find the passage that answers |
| Judge quality and origin | link analysis and PageRank; reliable information; reviews; original content | Read how pages link to each other, surface more authoritative sources, reward analysis with original research, show the original ahead of those who cite it |
| Block abuse | spam detection, with SpamBrain; exact match domain; removal-based demotion | Handle violations of the spam policies, avoid giving too much credit to domains built to match a query, demote sites with many valid removals |
| Arrange the results page | deduplication; site diversity; freshness; local news; crisis information | Avoid repeating similar results, generally show no more than two listings from one site in the top results, bring fresh content when the search calls for it, surface local sources, show help in a crisis |
One entry deserves a careful read. MUM is on the list, and the guide says it is not currently used for general ranking. It serves specific applications, such as searches for COVID-19 vaccine information. 3 Being on the list does not mean acting on every search.
Some entries already have their own piece on this site. BERT, described as a system for understanding meaning and intent, ties into the piece on search intent. RankBrain, which relates words to concepts, got its own piece. 3
Retired systems are still inside
At the end of the guide there is a section called “Retired systems.” It holds four famous names. 3
| System | What the guide records |
|---|---|
| Panda | Announced in 2011, to favor high-quality, original content; became part of the core ranking systems in 2015 |
| Penguin | Announced in 2012, against link spam; integrated into the core ranking systems in 2016 |
| Hummingbird | A major improvement to the overall ranking systems, made in August 2013 |
| Helpful content | Announced in 2022 as an update; in March 2024, it became part of the core ranking systems |
The section’s opening sentence explains the word. Retired systems were either folded into successor systems or made part of the core ranking systems. 3 As I read it, retired means the name left the list. The function carried on, inside another system.
Hummingbird has the most honest note on the page. After 2013, the ranking systems kept evolving, just as they had been evolving before. 3
Wrong dates circulate on the first page for this search in Brazil. One of the texts says Panda joined the core algorithm in 2016. 6 The guide says 2015. In 2016, the one that joined was Penguin. 3
A system and an update are different things
Back to the hospital. The lab works every day, without telling anyone. The wing under renovation has a sign on the door, with a start date and an expected end.
In Search, the department that works every day is the system. The renovation with a date has another name. This is called an update.
The guide separates the two in a single sentence. Systems are improved through rigorous testing and evaluation. Updates get a notice when that might be useful to content creators. 3 Who decides whether the notice is useful is Google. Not every change gets a notice.
Core update
Several times a year, Google makes broad changes to its search algorithms and systems. It calls these changes core updates, and it gives notice when they happen. 4
The same page has a detail few people read. There are also smaller core updates. Google does not announce them, because they are not widely noticeable. 4
A 2021 post on the Google blog gives the scale. In any given week, dozens of small, fully tested updates may go in. 7 Because there are so many, Google writes, sharing details about all of them is not useful. 7
Spam update
The systems that detect spam run all the time. Every so often, they get a notable improvement. That improvement is what Google calls a spam update. 8 The official page’s example is SpamBrain, Google’s AI-based spam-prevention system. 8
On the link spam update, the same page has a stark warning. When the systems neutralize spam links, the benefit those links gave is lost. According to Google, that benefit cannot be regained. 8
The system that stopped announcing
The clearest case of the difference is the reviews system. It is still on the guide’s active list. 3 On the status dashboard, its last update dates from November 2023. 9
In the record of that update, the text ends with a notice. The reviews system was now being improved at a regular and ongoing pace. So Google would no longer post its future updates on the dashboard. 9 The system keeps working. The notices stopped.
The status dashboard: the official record with dates
The Google Search Status Dashboard has a Ranking section. The dashboard’s help page says it shows the ranking updates relevant to website owners. 10
On September 14, 2026, the list started in November 2021 and had 41 entries. 9 I counted them by type. There are 16 core updates and 11 spam updates, one of them for links only. There are 7 reviews updates and 3 helpful content updates. Two issue notices, one page experience update, and one Discover update round out the count.
Read the list next to the guide, and two pieces fit together. After September 2023, the dashboard records no more helpful content updates. The guide explains why: in March 2024, that system became part of the core ranking systems. 3 9
The March 2024 core update is the longest on the list. It started on March 5 and ended on April 19, 45 days later. 9
How to read an update’s effect on your traffic is in the piece on why organic traffic dropped. It has Google’s own rules and the 2026 dates.
How Google tests a change before release
Think of the eye doctor. He flips the lens and asks: better, one or two? You don’t need to know optics to answer. You just compare the two images.
Google asks human raters a similar question. It shows them two sets of results for the same search, one with the proposed change and one without. Then it asks which results they prefer, and why. 5 This is called a side-by-side experiment.
The page sits on the How Search Works site and covers rigorous testing. It has the numbers for 2023. 5
| Type of test, in 2023 | Count |
|---|---|
| Launches | 4,781 |
| Live traffic experiments | 16,871 |
| Search quality tests | 719,326 |
| Side-by-side experiments | 124,942 |
By Google’s own count, that came to more than 700,000 experiments, which led to more than 4,000 improvements. 5 The next calculation is mine: for each launch, there were more than a hundred experiments.
A live traffic experiment starts small. The new feature is turned on for a tiny share of people, usually starting at 0.1%. That group is compared with a control group, which did not get the change. 5
The page lists some of the metrics in this test. Where people click, how many searches they run, whether they abandon a search, and how long they take to click. 5
Outside raters follow published guidelines. They assess whether content fulfills the search and rate the quality of the results. 5 The same page says these ratings do not directly impact ranking. They help Google benchmark the quality of its results. 5
The final decision goes through engineers and data scientists. According to Google, if a change cannot show that it makes things better for people, it does not launch. 5
For anyone with a site, the practical reading is simple. The update that reaches the dashboard went through these tests. So did the small change that never shows up there. 5 7
What the 2024 leak showed
Imagine a photo of an airplane cockpit. You can read the label on every switch. The photo does not show which switch the pilot flips, at what moment, or how hard.
In 2024, the documentation of an internal Google API, called Content Warehouse, leaked. According to Rand Fishkin, the material sat public on GitHub from March 27 to May 7, 2024. 11 Google confirmed to the press that the documents were real. 12
Mike King counted 2,596 modules and 14,014 described items. 13 Each item has a name and, sometimes, a line of explanation. This is called an attribute.
Names that match the guide
Some attributes repeat names the official guide uses. One field covers the exact match domain demotion, with the name of the guide’s system. 14 Others cover promoting and demoting product review pages. 14 And a field called babyPandaV2Demotion describes itself as a demotion applied on top of Panda. 14
The guide puts Panda among the retired systems and says it became part of the core in 2015. 3 In the documentation leaked in 2024, the name still labeled fields. The reading is mine: that fits what the guide calls incorporated.
The same module holds a field called siteAuthority. Its description gives no weight and no formula. 14 Two other fields, on a site’s focus on a topic, are explained in the piece on semantic silos.
Names the guide does not carry
By King’s count, the name NavBoost appears 84 times in the documentation, with five modules featuring it in the title. 13 The official guide has no entry under that name. 3
One of those modules stores counts of good clicks, bad clicks, and impressions. It splits them by country and language, and the documentation’s own examples include “US,” “BR,” and “pt-BR.” 15 On a public page, Google already writes that it uses aggregated and anonymized interaction data. That passage is in the piece on ranking first. 2
The guide’s list is partial, and the guide says so. 3 The leak shows, by name, that there is more outside it.
What it did not show
Back to the cockpit photo. The labels are legible. The flight manual was left out of the frame.
Fishkin wrote that the documentation does not show the weight of elements. Nor does it prove which ones the ranking systems use. 11 King wrote that nobody knows how the attributes are weighted in the scoring functions. Or whether everything there is used. 13 The Verge reported that the documents do not reveal how each element is weighted, if at all. 12
The material ages, too. The most recent date Fishkin found in the documents is August 2023. 11 And the documentation itself warns that fields in the clicks module may become legacy fields. 15 In its confirmation, Google cautioned against assumptions based on out-of-context, outdated, or incomplete information. 12
King proposes a general reading. For him, the picture of “the algorithm” as a single equation with weighted factors does not match the documentation. He describes a series of microservices, with perhaps more than a hundred ranking systems. 13 It is a hypothesis from someone who read the material, and he himself writes “may.”
Where the patent fits
The guide has a link almost nobody clicks. In the PageRank entry, it points to the original patent. 3 It is US 6,285,999 B1, by Lawrence Page, filed in the name of Stanford University. The application dates from January 9, 1998, and the grant from September 4, 2001. 16
In this case, Google confirms the use. PageRank remains among the core ranking systems. The same entry says how it works has evolved a lot since then. 3 The patent describes the 1998 method, and its term ended in 2018. 16 How PageRank calculates today, it does not tell you.
With most other patents, Google does not say whether it uses them. A patent does not prove implementation. It records a method someone wanted to protect.
What the first page says, and what it leaves out
On September 14, 2026, I read the first page for “algoritmo do google” in Brazil, in two readings. They were the top ten in Semrush’s Brazil database and the eight web results on Google in Brazil, live. 1 6 Five addresses were in both, and the total came to thirteen.
Two are Google’s own pages. One is news on another topic: a compression algorithm for AI models, created by Google researchers. 6 That leaves ten texts about search.
None of the ten cites the ranking systems guide. 6 Only one points to the status dashboard, in a link about major updates. None mentions the 2024 leak. 6 One cites human raters, and none describes the side-by-side experiment. 6
Three talk about 200 factors. One states the number as fact, and it is the source the AI Overview showed. Another calls it an estimate that circulates in the market. The third says Google never confirmed it. 1 6
The map on one sheet
The table translates what you hear into the document that checks it. The organization is mine, based on the sources above.
| What you hear | What it is | Where to check |
|---|---|---|
| “Google’s algorithm” | A set of systems; the guide lists seventeen and calls them some | Ranking systems guide 3 |
| “The algorithm changed” | An update with a name and a date, or one of the small ones, with no notice | Status dashboard and the core updates page 9 4 |
| “Panda got my site” | A retired name; the system has been part of the core since 2015 | The guide’s retired systems section 3 |
| “Google tested it first” | Live traffic experiments and side-by-side comparisons | The rigorous testing page 5 |
| “Factor X leaked” | An attribute that exists, with an unknown weight | The leaked documentation and its analyses 14 13 |
| “It’s in the patent” | A registered method, which does not prove use | The patent page 16 |
The question you ask starting today
From now on, when someone says the algorithm changed, you have two questions. Which system? And which update, with what date on the dashboard?
Without an answer to either, the claim is a guess. It may even be right, and it is still a guess.
Today, open the Google Search Status Dashboard and bookmark it. On a sheet, write down the last three core updates, with start and end dates. When traffic drops, that sheet comes in at the algorithm step, after you check the measurement, the site, and demand. The rest of this section is in semantic SEO.
Frequently asked questions
How many factors does Google’s algorithm use?
The current Google pages read for this piece give no number. 2 3 The guide lists seventeen systems and calls them some of the more notable ones. 3 The 14,014 items in the leak are stored attributes, with unknown weight. 13
What is the difference between a core update and a spam update?
A core update is a broad change to the search systems, made several times a year. 4 A spam update is a notable improvement to the spam detection systems, which run all the time. 8 Both appear on the status dashboard, with dates. 9
Does Google announce every update?
No. There are smaller core updates, with no announcement. 4 In a single week, dozens of small changes may go in. 7 And the reviews system stopped having its updates posted on the dashboard after November 2023. 9
Did the 2024 leak reveal the ranking factors?
It revealed the names of attributes Google stored. 13 The weight of each, and whether all of them are used, the documentation does not show. 11 12
Notes and sources
- Live reading of the Google in Brazil results page for “algoritmo do google” (Google algorithm), September 14, 2026, at https://www.google.com.br/search?q=algoritmo+do+google&hl=pt-BR&gl=br&pws=0, in the author’s own browser tab. The search ran on Google in Brazil, where we are based. The AI Overview (labeled “Visão geral criada por IA” in the Portuguese interface) opened with the sentence “O algoritmo de busca do Google é um sistema complexo que usa mais de 200 fatores para organizar e classificar as páginas da web e entregar o resultado mais útil ao usuário” (Google’s search algorithm is a complex system that uses more than 200 factors to organize and rank web pages and deliver the most useful result to the user), with “HubSpot +1” as its source. Web results, in order: Google (“Como o Google determina a classificação de resultados,” the Portuguese version of the page in note 2), Semrush, Shopify, Google for Developers (“Guia detalhado sobre como funciona a Busca do Google,” an in-depth guide to how Google Search works), Search Engine Journal, HubSpot, E-Dialog, and Conversion, with a video block between HubSpot and E-Dialog. Accessed September 14, 2026.
- Google, “How Does Google Determine Ranking Results” (page heading “Automatically generating and ranking results”), How Search Works site, https://www.google.com/search/howsearchworks/how-search-works/ranking-results/; Portuguese version, “Como o Google determina a classificação de resultados,” https://www.google.com/intl/pt-BR_ALL/search/howsearchworks/how-search-works/ranking-results/. Accessed September 14, 2026. Passages: “Search algorithms look at many factors and signals” and “We also use aggregated and anonymized interaction data to assess whether search results are relevant to queries.” A literal search for “200” in both versions: no occurrences.
- Google Search Central, “A guide to Google Search ranking systems,” https://developers.google.com/search/docs/appearance/ranking-systems-guide, page last updated December 10, 2025; Portuguese version, “Um guia sobre os sistemas de classificação da Pesquisa Google,” https://developers.google.com/search/docs/appearance/ranking-systems-guide?hl=pt-br, last updated December 18, 2025. Accessed September 14, 2026. Passages: “look at many factors and signals about hundreds of billions of web pages and other content in our Search index […] all in a fraction of a second”; “This page is a guide to understanding some of our more notable ranking systems”; “core ranking systems, which are the underlying technologies that produce search results in response to queries”; and “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”. The seventeen active entries, in the page’s order: BERT; crisis information systems; deduplication systems; exact match domain system; freshness systems; link analysis systems and PageRank; local news systems; MUM; neural matching; original content systems; removal-based demotion systems; passage ranking system; RankBrain; reliable information systems; reviews system; site diversity system; spam detection systems. On BERT, “understand how combinations of words express different meanings and intent”; on RankBrain, “helps us understand how words are related to concepts”; on MUM, “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”. On PageRank, “How PageRank works has evolved a lot since then, and it continues to be part of our core ranking systems”, with a link to https://patents.google.com/patent/US6285999. The “Retired systems” section: “They’ve either been incorporated into successor systems or made part of our core ranking systems”; Panda announced in 2011 and part of the core ranking systems in 2015; Penguin announced in 2012 and integrated in 2016; Hummingbird, “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”; the helpful content system announced in 2022 and part of the core ranking systems in March 2024. A literal search for “NavBoost” and for “200”: no occurrences. The grouping by job in the table is the author’s.
- Google Search Central, “Google Search’s core updates and your website,” https://developers.google.com/search/docs/appearance/core-updates, page last updated December 10, 2025. Accessed September 14, 2026. Passages: “Several times a year, Google makes significant, broad changes to our search algorithms and systems. We refer to these as core updates, and we give notice when they happen” and “We’re continually making updates to our search algorithms, including smaller core updates. These updates are not announced because they aren’t widely noticeable”. A literal search for “200”: no occurrences.
- Google, “Improving Search with rigorous testing,” How Search Works site, https://www.google.com/search/howsearchworks/how-search-works/rigorous-testing/; Portuguese version, “Melhorando a Pesquisa com testes rigorosos,” https://www.google.com/intl/pt-BR/search/howsearchworks/how-search-works/rigorous-testing/. Undated page, with 2023 figures. Accessed September 14, 2026. Passages: “In 2023, we ran over 700,000 experiments that resulted in more than 4,000 improvements to Search”; “4,781 launches”, “16,871 live traffic experiments”, “719,326 search quality tests”, and “124,942 side-by-side experiments”; “usually starting at 0.1%. We then compare the experiment group to a control group”; the metrics “what people click on, how many queries were done, whether queries were abandoned, or how long it took for people to click on a result”; “Raters assess how well website content fulfills a search request”; “These ratings do not directly impact ranking”; “our most experienced engineers and data scientists”; “If we can’t show that a change actually makes things better for people, we don’t launch it”; and “we show Raters two different sets of Search results: one with the proposed change already implemented and one without. We ask them which results they prefer and why.” The Portuguese version calls them “experimentos de comparação” (comparison experiments), with the same 124,942. Dividing more than 700,000 experiments by 4,781 launches gives more than 146; the division is the author’s. A literal search for “200”: no occurrences.
- SEO Genome survey, September 14, 2026. Top ten organic results in Semrush’s Brazil database for “algoritmo do google”: google.com (How Search Works), shopify.com, br.hubspot.com, braziljournal.com, edialog.com.br, pt.semrush.com, bbc.com, cmlo.co, hostgator.com.br, and lemostelecom.com.br. Added to the eight in note 1, with five in common, they came to thirteen addresses, all downloaded and read the same day. All but the Search Engine Journal article, which Google showed translated, are in Portuguese. Two are Google’s. The one at https://braziljournal.com/turboquant-o-novo-algoritmo-do-google-derruba-empresas-de-chips-de-memoria/ covers TurboQuant, a compression technique for AI models. The ten about search: https://www.shopify.com/br/blog/algoritmo-do-google, https://br.hubspot.com/blog/marketing/algoritmo-do-google, https://www.edialog.com.br/algoritmo-do-google/, https://pt.semrush.com/blog/algoritmo-do-google/, https://www.bbc.com/portuguese/articles/c0wwz4y49gro, https://cmlo.co/seo/algoritmo-do-google/, https://www.hostgator.com.br/blog/algoritmo-google-quais-sao/, https://www.lemostelecom.com.br/noticias-detalhe/680/como-funciona-o-algoritmo-dos-buscadores-como-o-google-, https://www.searchenginejournal.com/google-algorithm-history/, and https://www.conversion.com.br/blog/atualizacoes-algoritmo-google/. None links to or mentions the ranking systems guide; only Shopify’s links to the status dashboard; none mentions the 2024 leak; only Semrush’s mentions outside raters; none describes the side-by-side experiment. On the 200 factors: HubSpot, “O algoritmo de busca do Google é um sistema extremamente sofisticado que utiliza mais de 200 fatores para classificar os resultados” (Google’s search algorithm is an extremely sophisticated system that uses more than 200 factors to rank results); Semrush, “circula no mercado digital uma estimativa de pelo menos 200 fatores” (an estimate of at least 200 factors circulates in the digital market); HostGator, “Atualmente supõe-se que são mais de 200 fatores, mas essa informação nunca foi confirmada pelo Google” (Today it is assumed there are more than 200 factors, but Google has never confirmed that). E-Dialog: “Em 2016, o Panda foi incorporado ao algoritmo principal do Google” (In 2016, Panda was incorporated into Google’s core algorithm). Accessed September 14, 2026.
- Danny Sullivan, “How Google updates Search to improve our results,” the official Google blog, June 2, 2021, https://blog.google/products/search/how-we-update-search-improve-results/ (redirects to the blog’s new address). Accessed September 14, 2026. Passages: “In any given week, we might implement dozens of updates that are meant to improve Search in incremental ways. These are improvements that have been fully tested and evaluated through our rating process” and “Because there are so many incremental updates, it’s not useful for us to share details about all of them.”
- Google Search Central, “Google Search spam updates and your site,” https://developers.google.com/search/docs/appearance/spam-updates, page last updated December 10, 2025. Accessed September 14, 2026. Passages: “While Google’s automated systems to detect search spam are constantly operating, we occasionally make notable improvements to how they work. When we do, we refer to this as a spam update”; “SpamBrain is our AI-based spam-prevention system”; and, on the link spam update, “any ranking benefit the links may have previously generated for your site is lost. Any potential ranking benefits generated by those links cannot be regained.”
- Google Search Status Dashboard, “History for Ranking,” https://status.search.google.com/products/rGHU1u87FJnkP6W2GwMi/history. Accessed September 14, 2026. The author’s count: 41 entries, from November 3, 2021, to August 18, 2026, made up of 16 core updates, 11 spam updates (one of them the December 2022 link spam update), 7 reviews and product reviews updates (the last on November 8, 2023), 3 helpful content updates (August 2022, December 2022, and September 2023), 2 notices reading “Ranking is experiencing an ongoing issue,” the Page experience update for desktop, and the February 2026 Discover update. Times in the US Pacific time zone. The record of the November 2023 reviews update, https://status.search.google.com/incidents/VqRTcmyQCwfpuYeaVNfn, ends with “Going forward, as the reviews system is now being improved at a regular and ongoing pace, we will no longer be posting about future updates on the status dashboard.” The record of the March 2024 core update, https://status.search.google.com/incidents/1jW2F89qC2NxJBWGiKxE, runs from March 5 to April 19, 2024, a duration of 45 days, the longest on the list.
- Google Search Central, “Using the Google Search Status Dashboard,” https://developers.google.com/search/help/status-dashboard. Accessed September 14, 2026. Passage: “The dashboard also shows the latest ranking updates made to Google Search that are relevant to website owners.”
- Rand Fishkin, “An Anonymous Source Shared Thousands of Leaked Google Search API Documents with Me; Everyone in SEO Should See Them,” SparkToro, published May 27, 2024, US time (May 28 in UTC), https://sparktoro.com/blog/an-anonymous-source-shared-thousands-of-leaked-google-search-api-documents-with-me-everyone-in-seo-should-see-them/. Accessed September 14, 2026. Passages: “this code was uploaded to GitHub on Mar 27, 2024 and not removed until May 7, 2024”; “This documentation doesn’t show things like the weight of particular elements in the search ranking algorithm, nor does it prove which elements are used in the ranking systems”; and “The most recent date I can find referenced in the API docs is August of 2023”.
- Mia Sato, “Google confirms the leaked Search documents are real,” The Verge, May 29, 2024, https://www.theverge.com/2024/5/29/24167407/google-search-algorithm-documents-leak-confirmation. Accessed September 14, 2026. Passages: from Google spokesperson Davis Thompson, “We would caution against making inaccurate assumptions about Search based on out-of-context, outdated, or incomplete information”; and “The documents also do not reveal how different elements are weighted in search, if at all.”
- Mike King, “Secrets from the Google Algorithm Leak: Search’s Internal Engineering Documentation and What it Means,” iPullRank, May 27, 2024, https://ipullrank.com/google-algo-leak. Accessed September 14, 2026. Passages: “There are 2,596 modules represented in the API documentation with 14,014 attributes (features)”; “We do not know how features are weighted in the various downstream scoring functions. We don’t know if everything available is being used”; “The leaked documentation specifies ‘Navboost’ by name 84 times with five modules featuring Navboost in the title”; and “Conceptually, you may think of ‘the Google algorithm’ as one thing, a giant equation with a series of weighted ranking factors. In reality, it’s a series of microservices […] there may be over a hundred different ranking systems.”
- Documentation of Google’s Content Warehouse API, leaked in 2024, module
CompressedQualitySignals, public mirror of thegoogle_api_content_warehousepackage v0.4.0, https://google-api-content-warehouse.hexdocs.pm/0.4.0/GoogleApi.ContentWarehouse.V1.Model.CompressedQualitySignals.html. Accessed September 14, 2026. Attributes:exactMatchDomainDemotion, “exact_match_domain_demotion: converted from QualityBoost.emd.boost”;productReviewPDemoteSite, “Product review demotion/promotion confidences”, next toproductReviewPPromotePageandproductReviewPDemotePage;babyPandaV2Demotion, “New BabyPanda demotion, applied on top of Panda”; andsiteAuthority, “site_authority: converted from quality_nsr.SiteAuthority, applied in Qstar”. No description gives a weight. The link between these names and the guide’s entries in note 3 is the author’s reading. - Documentation of Google’s Content Warehouse API, leaked in 2024, module
QualityNavboostCrapsCrapsData, same mirror, https://google-api-content-warehouse.hexdocs.pm/0.4.0/GoogleApi.ContentWarehouse.V1.Model.QualityNavboostCrapsCrapsData.html. Accessed September 14, 2026. AttributesgoodClicks,badClicks,impressions,lastLongestClicks,country(“Examples: “US”, “FR”, “BR""), andlanguage(“Examples: “en”, “fr”, “pt-BR""). Onimpressions: “These fields may become legacy fields; we may retire them and use the squashed field (below) instead”. - Patent US 6,285,999 B1, “Method for node ranking in a linked database,” inventor Lawrence Page, application filed by Leland Stanford Junior University, https://patents.google.com/patent/US6285999B1/en. Accessed September 14, 2026. Priority date January 10, 1997; application filed January 9, 1998; granted September 4, 2001; an assignment record to Google LLC, a change of name from Google Inc., dated December 5, 2017; status “Expired - Lifetime”, with anticipated expiration on January 9, 2018, and the platform’s caveat that the status is an assumption. The patent records the 1998 method; current use and the evolution of PageRank come from the guide, not from the patent.
Want to see SEO Genome more often on Google? Add it as a preferred source.
Let us look at what the machine understands about your company
A diagnostic conversation, with no slide deck. You bring your website address. We bring what Google and the AI assistants already know about it today.