Google Algorithm Updates Explained: Core, Helpful Content, Spam and Reviews — What Each One Actually Targets

There is no single Google algorithm. Google runs many automated ranking systems over hundreds of billions of web pages, and a “Google algorithm update” is a change to one or more of those systems. The updates that matter to site owners fall into a few named families: core updates, which broadly reassess how content measures up; spam updates, which improve the systems that catch policy violations; and the reviews system, which rewards genuinely useful reviews. The old helpful content system, once a standalone update of its own, was folded into the core ranking systems in March 2024. This guide explains each family in plain language — what it targets, who it affects, and what Google itself says to do about it — using only Google’s own documentation and its Search Status Dashboard. Checked 11 November 2026. One update is rolling out right now: the November 2026 core update began on 10 November 2026 and was still rolling out when this article was checked. Google has said nothing about what it changes, and this guide does not guess.

The Short Answer: What a Google Algorithm Update Actually Is

<img src="

” alt=”A winding road splitting into three blank signposts at a junction, soft morning light, no words on the signs”>

When people say “the Google algorithm”, they usually picture one machine with one set of rules. That picture is wrong, and most of the confusion about updates starts there. Google’s own guide to its ranking systems describes many automated systems working together over hundreds of billions of web pages, presenting results in a fraction of a second. Some systems understand language. Some judge freshness. Some weigh links. Some catch spam. An “update” is a change to one or more of those systems — sometimes a small, unannounced tweak, sometimes a large, announced rollout that Google lists on its Search Status Dashboard.

That is why updates feel so uneven. A change to the spam-detection systems hits sites that break spam policies and leaves everyone else alone. A core update, by contrast, is a broad reassessment: Google changes how its systems weigh content in general, and pages move up or down because they were reassessed against everything else, not because they were caught doing something wrong. A page that drops after a core update has not necessarily done anything wrong at all. Other pages may simply have been judged more helpful this time around.

The three families in this guide’s title cover almost everything a site owner will ever need to understand. Core updates are the big, broad reassessments, announced several times a year. Spam updates are notable improvements to the spam-detection systems, which run constantly in the background. The reviews system rewards review content that shows real research and first-hand experience. The helpful content story belongs with the core updates now: the system Google announced in 2022 was folded into core ranking in March 2024, and its guidance — write for people, not for search engines — survives as the self-assessment every site owner should still use.

One thing is rolling out as you read this. Google’s Search Status Dashboard shows the November 2026 core update began on 10 November 2026 and was still rolling out at the check for this article on 11 November 2026. Google’s entire announcement is one sentence: the update applies globally and to all languages, and the rollout may take up to 2 weeks to complete. Google has said nothing about what it changes, which systems it touches, or who it will affect — so this guide says nothing about that either. If your rankings move while it is rolling out, the honest position is that you do not yet know why, and neither does anyone outside Google. The dashboard is the only authoritative place to check whether it has finished; this guide’s final section tells you exactly what to do while you wait. If you want to control what you see in search results while updates roll out, our guide to turning off AI on Google Search covers the controls on your side of the screen.

The rest of this guide works through the system family by family. It starts with how ranking works, then explains core updates, the helpful content story, spam updates, the sixteen spam policies, the site-reputation rule and its European exception, the reviews system, and what Google actually says about AI-generated content. Then comes the timeline of every confirmed update from March 2024 to today, Google’s own method for telling whether an update hit your site, what recovery really looks like, the myths that cost site owners the most time, and a checklist for this week.

How Google Ranking Works: Systems, Not One Algorithm

Google’s ranking systems are designed to sort hundreds of billions of web pages and present the most relevant, useful results in a fraction of a second. Ranking is designed to work at page level: each page is judged largely on its own merits. Site-wide signals also count, but Google is explicit about their limits: “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.” A strong site can publish a weak page, and a weak site can publish a strong one.

The named systems are easier to understand than their names suggest. BERT helps Google understand how combinations of words express different meanings and intent — the same words in a different order can mean different things, and BERT is built for that. RankBrain helps Google understand how words relate to concepts, so it can return relevant results even for a query whose exact words do not appear on the page. Neural matching understands representations of concepts in queries and pages and matches them to one another. PageRank, one of Google’s original core ranking systems, analyses links — how PageRank works has evolved a lot since Google first launched, and it continues to be part of the core ranking systems, but it is one system among many now, not the algorithm itself. For readers who want the deeper explanation of how these AI systems work, our guide to how AI works covers the underlying ideas, and our BERT, Llama, Whisper and RAG explainer goes into BERT itself in depth.

Other systems handle different jobs. Freshness systems are designed to show fresher content for queries where it would be expected — a search for a match happening now is not the same as a search for a historical event. Original content systems help ensure original content, including original reporting, is shown prominently ahead of pages that merely cite it. The site diversity system means Google generally will not show more than two web page listings from the same site in its top results. Deduplication systems stop a page that appears as a featured snippet from being repeated later on the first page. The exact match domain system ensures Google does not give too much credit for content hosted under domains designed to exactly match particular queries — owning the domain that matches the search is not, by itself, a ranking win.

Then there are the systems most people never hear about: passage ranking, crisis information systems, local news systems, reliable information systems, removal-based demotion systems, and spam detection systems including SpamBrain. MUM, Google’s language model capable of both understanding and generating language, is not currently used for general ranking in Search; Google uses it for specific applications such as improving searches for COVID-19 vaccine information and improving featured snippet callouts.

Some famous names from search history are no longer standalone systems at all. Panda, announced in 2011, became part of the core ranking systems in 2015. Penguin, announced in 2012, was integrated into the core ranking systems in 2016. Hummingbird dates to August 2013. And the helpful content system, announced in 2022, evolved in March 2024 and became part of the core ranking systems. The names retired; the work they did did not disappear — it was absorbed into core ranking. That is the pattern to remember throughout this guide: Google’s systems consolidate over time, and the labels change faster than the underlying goals.

The practical consequence is simple. There is no single dial to turn and no single score to raise. A page ranks because many systems, each judging a different quality, agree that it is useful for a particular query. An update changes one or more of those systems, and the pages that move are the ones whose strengths or weaknesses line up with what changed. That is why the same update can help one site, hurt another, and leave a third untouched — and why chasing a single supposed ranking factor is a poor use of time.

Core Updates: The Broad Reassessment

Core updates are the updates most site owners worry about, and the ones most often misunderstood. Google says they happen several times a year and are “significant, broad changes to our search algorithms and systems”. Broad is the key word: core updates “don’t target specific sites or individual web pages”. They are not penalties. They are not aimed at anyone. They are a reassessment of how content measures up, across the board.

Google explains this with an analogy that is worth quoting in full, because it clears up more confusion than any technical description. Imagine you made a list of the top 100 restaurants in a city in 2021. A few years later you refresh the list. Some new restaurants that did not exist before will now be included. Some restaurants that were previously on the list will move down, or drop off entirely. The restaurants that moved down have not necessarily got worse — the list changed because new places appeared and the whole field was reassessed. Pages that move down after a core update are in the same position: they “aren’t necessarily ‘bad'”. They were reassessed against a changed field.

That is why a good page can drop and a page can drop even when its owner changed nothing. The content did not get worse; the competition got better, or Google’s systems got better at telling the difference. It is also why there is no such thing as “fixing” a core update hit in the way you fix a penalty. There is no violation to remove and no reconsideration request to file. The only response Google offers is the one it has offered since the helpful content work began: make the content genuinely more helpful for the people reading it, and be patient.

There is one more feature of core updates that surprises people: the big, announced ones are not the only ones. Google says it is “continually making updates to our search algorithms, including smaller core updates. These updates are not announced because they aren’t widely noticeable.” That matters for recovery, and this guide returns to it in the recovery section: improvements to your content can show effect without waiting for the next major core update, because smaller updates run all the time. The announced core update is the visible peak of continuous change, not the whole mountain.

Core updates are announced on the Search Status Dashboard, with a start date and, when finished, a completion message. The dashboard is the only authoritative source for those dates — and the timeline section of this guide is built entirely from it. If a blog post, a forum thread, or a tool tells you a core update is happening and the dashboard says nothing, the dashboard wins. Smaller core updates are not announced at all, which means some ranking movement will never have a named update attached to it. That is normal, and it is one reason the guide’s section on diagnosing a drop starts with checking the dashboard dates rather than assuming the nearest update is the cause.

The Helpful Content Story: From Standalone System to Core

The phrase “helpful content update” still appears in search advice everywhere, but the thing it named no longer exists as a standalone system. Google announced the helpful content system in 2022. In March 2024, it evolved and became part of the core ranking systems. Google’s own words on the change are that “there’s no longer one signal or system” assessing helpfulness. Helpfulness is now assessed by the core systems using a variety of signals and approaches — which means there has been no standalone “helpful content update” since March 2024, and any advice that treats helpful content as a separate system with its own rollout dates is out of date.

What survived the change is the guidance, and it is the most useful material Google has ever published for site owners. Google’s automated ranking systems are designed to prioritise “helpful, reliable information that’s created to benefit people, and not content that’s created to manipulate search engine rankings”. The test Google offers is a self-assessment, and it is worth running honestly on your own pages. The content and quality questions ask whether a page provides original information, reporting, research or analysis; a substantial, complete, comprehensive description of its topic; insightful analysis or interesting information beyond the obvious; something that avoids simply copying or rewriting sources; substantial value when compared with other pages in the search results; and work that is produced well rather than sloppily or hastily. It also asks whether the content is mass-produced across a network of sites — if it is, that is a warning sign in itself.

The expertise questions ask whether the content has clear sourcing and evidence of the expertise behind it, background information about the author or the site, and whether it was written or reviewed by an expert or enthusiast who demonstrably knows the topic. They also ask whether the content is free of easily-verified factual errors. These questions are the written form of E-E-A-T: experience, expertise, authoritativeness, and trustworthiness. Of the four, Google says trust is the most important. And E-E-A-T itself is not a specific ranking factor — it is a framework for what the ranking systems are trying to reward. For topics that affect health, financial stability, safety, or the welfare of society — what Google calls YMYL topics — the systems give even more weight to strong E-E-A-T, and the content “must be highly accurate and consistent with established expert consensus”.

Then there is the people-first test. Ask whether the content has an existing or intended audience who would find it useful if they came directly to it; whether it demonstrates first-hand expertise and depth of knowledge, including where that expertise is visible; whether the site has a primary purpose or focus; whether a reader will leave feeling they have learned enough about the topic to achieve their goal; and whether the reader will leave feeling they had a satisfying experience. If the honest answer to most of those is no, the page is not people-first, whatever its keyword targeting looks like.

Google also lists the warning signs of search-engine-first content, and they read like a catalogue of the shortcuts that fill the web. Content made primarily to attract visits from search engines. Lots of content on many different topics in the hope that some of it performs well. Extensive use of automation to produce content on many topics. Content that mainly summarises what others have said without adding much value. Writing about things simply because they seem trending, not because you would write about them otherwise. Content that leaves readers needing to search again to get better information. Content written to a particular word count because someone heard Google has a preferred word count — Google’s answer is blunt: “No, we don’t.” Entering a niche without real expertise because it seems likely to bring traffic. Content that promises an answer when there is no confirmed answer. Changing the dates of pages to make them seem fresh, or adding or removing large amounts of content primarily to make pages seem “fresh” — Google’s answer to that one is equally blunt: “No, it won’t” help.

Finally, the Who, How and Why test. Who created the content — Google strongly encourages adding accurate authorship information, and warns against deceptive authorship, including fabricating creator profiles with AI-generated headshots, made-up names, or false credentials, which it calls a form of deception. How the content was created — disclose AI or automation where a reader would reasonably ask “How was this created?” Why it was created — to help people, not to attract visits. Quality raters, the people who assess whether Google’s algorithms seem to be providing good results, feed into this picture too, giving Google insight into whether the systems are working. And page experience counts: Google’s core ranking systems look to reward content that provides a good page experience.

The helpful content story, in one sentence, is this: the system became part of core ranking, but the standard it set is still the standard. The March 2024 change did not lower the bar or move it somewhere else. It made helpfulness a property of the core systems themselves, judged by many signals at once. The self-assessment questions above are still the best tool a site owner has for judging their own pages — and they are the tool this guide’s recovery section uses.

Spam Updates: What They Target and How SpamBrain Works

Spam updates are different from core updates in a way that matters. Google’s automated systems to detect search spam are constantly operating — they do not switch on for an update and off again afterwards. A spam update is what Google calls it when it makes a notable improvement to how those systems work, and it shares the news on its list of Google Search ranking updates. The systems run all the time; the update is a step change in how well they work.

SpamBrain is the best-known of those systems: Google describes it as “our AI-based spam-prevention system”. It is one of a variety of systems that analyse patterns and signals to identify spam content, however it is produced — by hand, by machine, or by a mix of the two. The point of naming it is not the technology. It is the scope: spam detection is pattern detection, not a search for a single trick. A page that looks like spam in its structure, its links, its scale, or its behaviour can be caught even if no individual sentence breaks a rule.

What happens to a site that violates the spam policies is stated plainly: “Sites that violate our policies may rank lower in results or not appear in results at all.” That is the whole range, from a demotion to disappearance. And the way back is slow by design. Making changes may help a site improve “if our automated systems learn over a period of months that the site complies with our spam policies”. Months, not days — the systems need time to relearn that a site has changed, and Google offers no shortcut around that.

This is why a spam update feels different from a core update. A core update is a reassessment: nothing was necessarily done wrong. A spam update means a policy line was crossed. The fix is not “make the content a bit better” but “stop violating the policy, clean up what was done, and wait for the systems to notice”. The recovery section of this guide separates the two paths for exactly that reason — they are different problems with different timelines, and treating a spam problem as a core-update problem wastes the months that matter.

There is one hard edge that belongs to spam alone. When Google’s systems remove the effects of spammy links, “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.” No other spam type has this rule. Fix a doorway page and the page can recover; remove spammy links and the rankings they bought are gone for good. Link spam is the one violation whose gains cannot be earned back by cleaning up. That fact shapes the whole link-spam policy, and it is the reason the next section treats link spam as seriously as it does.

Spam updates in 2026 have come in a steady sequence, and the timeline section of this guide lists them with their verified dates: March, June, August, and September, with the September update running longer than the others. Each was announced on the dashboard as a global update applying to all languages. None of them came with impact figures, and Google publishes none — a point this guide repeats wherever it matters, because the absence of numbers is itself a fact worth knowing.

The Spam Policies, One by One

Google’s spam policies are the rules behind spam updates: sixteen policies covering the techniques used to deceive users or manipulate search systems into featuring content prominently — or to manipulate generative AI responses in search. The policies apply to all web search results, including results from Google’s own properties. Detection happens both through automated systems and, as needed, through human review that can result in a manual action. This section tours all sixteen, grouped for readability, with Google’s own definitions and one illustrative example for each group. The full policies are linked in the Sources section.

Content tricks. Cloaking is showing different content to users than to Google — the search engine sees one page and the visitor sees another. Doorway abuse is creating pages or sites designed to rank for specific, similar searches while funnelling users to a single destination, so the search results are padded with near-identical entries that all lead to the same place. Hidden text and link abuse is hiding text or links where users cannot see them — in tiny type, in matching colours, behind images — while search engines can. Keyword stuffing is loading a page with keywords or numbers in an attempt to manipulate rankings, repeating the same terms until the page reads like a list rather than writing. Sneaky redirects send users to a different destination than the search engine was shown, or than the result promised. Machine-generated traffic is sending automated queries or requests to Google in an attempt to manipulate results or to scrape them.

Scale tricks. Scaled content abuse is “when many pages are generated for the primary purpose of manipulating search rankings and not helping users” — and Google is explicit that it applies “no matter how it’s created”: generative AI, scraping, synonymising or translating existing content, stitching pages together, or running multiple sites to hide the scale. The production method is irrelevant; the purpose and the scale are the violation. Scraping is taking content from other sites and republishing it without adding value, so that search results fill with copies of the same material. Thin affiliation is the affiliate version of the same idea: product descriptions or reviews copied directly from the original merchant without any original content or added value. A good affiliate page, by contrast, adds price information, original reviews, testing and ratings, navigation, and comparisons — work the merchant did not do.

Authority tricks. Expired domain abuse is buying an expired domain and repurposing it primarily to manipulate search rankings by hosting content that provides little to no value to users. Google’s own examples make the shape of it clear: affiliate content on a former government-agency site, medical products on a former charity site, casino content on a former elementary-school site. The domain’s old reputation is being spent on content that has nothing to do with it. Link spam includes buying or selling links for ranking purposes — but there is an important carve-out: “It’s not a violation of our policies to have such links as long as they are qualified with a rel=’nofollow’ or rel=’sponsored’ attribute.” A paid link that is properly labelled is not link spam; a paid link passed off as an editorial vote is. Site reputation abuse is covered in its own section below, because it has its own history and its own European exception. Google’s own spokesperson has warned publicly that AI spam tactics will catch up with the sites using them — our report on Google’s Mueller warning about AI spam tricks covers that warning, and it sits squarely on the scaled-content-abuse policy described here.

User and security problems. Hacked content is content placed on a site by a third party without the owner’s permission, to host spam or malicious pages — the site owner is a victim, but the content is still spam in the results. User-generated spam is spam content posted by visitors: in comments, forum posts, profiles, or any other place a site lets users publish. Google has extended its user-generated spam guidance to AI tools, and our report on that extension explains what changed for sites that let users — or their tools — post. Malicious practices covers content designed to harm users: malware, phishing, and other deceptive software. Misleading functionality is content that promises one thing and does another — a page that claims to be a tool, a download, or a service and delivers something else entirely.

Beyond the sixteen policies sit the demotion and removal practices. Legal removals: a significant volume of valid copyright removal requests can demote other content from a site, and the same applies to defamation, counterfeit goods, and court orders. Personal information removals cover content that exposes private details. Policy circumvention — creating new subdomains, subdirectories, or sites to keep violating a policy after enforcement — can lead to Google removing more sections of a site from search results. Scam and fraud is what it sounds like: content designed to defraud users. And material depicting the sexual abuse of children is always removed, with sites carrying a significant proportion of such content demoted.

Two things are worth noticing about the list as a whole. First, it is a list of behaviours, not of technologies. AI appears in it only as a production method for scaled content abuse, not as a violation in itself — a distinction the AI section of this guide takes up in full. Second, the policies are enforced by a mix of automation and human review, which means a site can be caught by a system, by a person, or by both, and the remedy — fix the violation, wait for the systems to learn over months — is the same either way.

Site Reputation Abuse: The “Parasite SEO” Rule — and the Europe Exception

Site reputation abuse is the policy most people know by its nickname: “parasite SEO”. The idea is simple. A third party publishes pages on a host site mainly to exploit the host’s already-established ranking signals — the host’s reputation, in other words — rather than because the content belongs there. The pages are published with little or no first-party oversight or involvement, and their purpose is to manipulate search rankings by taking advantage of the first-party site’s ranking signals. The policy was announced in March 2024 alongside two others, and it took effect on 5 May 2024.

The policy does not treat all third-party content as a violation — only content hosted without close oversight and intended to manipulate rankings. Google names the kinds of third-party content that are not inconsistent with the policy: wire services, syndicated news, forums and comments, opinion columns, and advertorial or native advertising aimed at regular readers rather than rankings. A coupons area is fine when the host is actively involved — sourcing coupons directly and taking editorial responsibility for them. White-label redistribution of coupons primarily to manipulate rankings is not. The difference is oversight and intent, not the mere presence of someone else’s content on your site.

When a human reviewer looks at a case, they weigh several factors: how the content is presented, its quality, the stated authorship, and whether identical content appears on multiple sites. Unauthored, unintegrated third-party pages that appear in many places at once are the shape of a violation; content that is presented as part of the host site, written to its standard, and owned by someone identifiable is the shape of a legitimate arrangement. The test is practical, not formal: does the host site actually stand behind this content, or is it renting out its reputation?

Then there is the exception, and it is a large one. The consequences of site reputation abuse differ by location. Outside the European Economic Area, pages that violate the policy “may be subject to a manual action” — a human reviewer can demote them. Inside the EEA, the pages “may be categorized as separate from the main domain but won’t be subject to the impact of manual action” — they “will rank independently… on their own merits”, with no manual-action demotion. The same pages, the same policy, two different outcomes depending on where the site is. The regulatory backdrop is the European framework that treats search results differently from the rest of the world; for that context, our guide to AI regulation around the world covers the EU AI Act, US rules, and China’s approach. And the change itself was news on this site: our report on Google dropping site-reputation penalties across Europe covers the EEA change as it happened.

Affected sites are notified in the Manual actions report in Search Console and in the Search Console message centre, and they can appeal through a reconsideration request — EEA sites can also use mediation. That notification route is worth remembering: if your site is ever hit by a manual action of any kind, Search Console is where Google tells you. The site-reputation policy is also a good illustration of a wider truth about Google’s rules: they are enforced by a mix of systems and people, and the same violation can look different in different places. A site owner with third-party content on their site should ask the oversight question honestly — who wrote this, who checked it, and would we publish it if it brought no rankings at all? — because that is the question Google’s reviewers are asking.

The Reviews System: What Google Rewards in Reviews

The reviews system has one job: to better reward high quality reviews. Google defines those as content that “provides insightful analysis and original research and is written by experts or enthusiasts who know the topic well”. The system works to ensure people see reviews that share in-depth research, “rather than thin content that simply summarizes a bunch of products, services or other things”. A review that restates the manufacturer’s specification sheet and adds a star rating is the thin content the system is built to pass over. A review that tests the thing, measures it, compares it, and explains what it is actually like to use is the content the system is built to reward.

What the system evaluates is specific. It evaluates articles, blog posts, pages or similar first-party standalone content written with the purpose of providing a recommendation, giving an opinion, or providing analysis. That covers single reviews, head-to-head comparisons, and ranked lists, about products such as laptops and winter jackets, media such as movies and video games, or services and businesses such as restaurants and fashion brands. What it does not evaluate is third-party reviews, such as those posted by users in the reviews section of a product or services page. Your customers’ star ratings on your product page are not what this system judges; your own review article is.

Evaluation is primarily at page level. But for sites that have a substantial amount of review content, any content within the site might be evaluated by the system — a review-heavy site carries its review quality across its whole domain. The system covers eleven languages: English, Spanish, German, French, Italian, Vietnamese, Indonesian, Russian, Dutch, Portuguese, and Polish. And it is “improved at a regular and ongoing pace” — there is no single reviews update to watch for, because the system is updated continuously.

Recovery works the way recovery works everywhere else in this guide: content affected by the reviews system “may recover over time, if you’ve made improvements to your content”. But Google adds a caution that belongs in every recovery conversation: “our automated assessment of review content is only one of many factors used in ranking content.” Improving a review helps; it does not guarantee a return to a previous position, because ranking is the product of many systems, not one.

Google also publishes its own best practices for writing reviews, and they are the most concrete writing advice in this whole guide. Evaluate from a user’s perspective. Demonstrate that you are knowledgeable about what you are reviewing. Provide evidence of your own experience, such as visuals, audio, or other links. Share quantitative measurements about how something measures up in various categories of performance. Explain what sets something apart from its competitors. Cover comparable things to consider. Discuss the benefits and drawbacks of something, based on your own original research. Describe how a product has evolved from previous models or releases. Focus on the most important decision-making factors. Describe key choices in how a product has been designed and their effect on users, beyond what the manufacturer says. Include links to other useful resources, and consider linking to multiple sellers. When you recommend something as the best overall, include why you consider it the best, with first-hand supporting evidence. Ranked lists must have enough useful content to stand on their own — a list that only works as a set of links to other pages is not a review.

Two further notes round out the picture. Reviews often use affiliate links, and Google points reviewers to its own position on affiliate programs — the thin affiliation policy in the previous section is the rule that applies. And when writing reviews, Google says to focus on the quality and originality of the reviews, not the length. A short review built on real testing beats a long review built on the specification sheet, and no word count will rescue a review that never touched the product.

AI Content and the Algorithm: What Google Actually Says

No question about Google’s algorithm produces more confident wrong answers than this one: is AI-generated content banned? The answer, from Google’s own guidance published in February 2023, is no. “Google’s ranking systems aim to reward original, high-quality content that demonstrates qualities of what we call E-E-A-T” — and the focus, Google says, is “on the quality of content, rather than how content is produced”. That focus, it adds, “has helped us deliver reliable, high quality results to users for years”. How content was produced is not the test. Whether it is original, high-quality, and demonstrably expert is.

The line is drawn at purpose. “Using automation — including AI — to generate content with the primary purpose of manipulating ranking in search results is a violation of our spam policies.” The same tool, the same model, the same output can be legitimate or spam depending on why it was made. Google’s own examples of legitimate automation are deliberately ordinary: sports scores, weather forecasts, and transcripts. Automation “has long been used to generate helpful content”, and “not all use of automation, including AI generation, is spam”. The question to ask of any AI-assisted page is not “was a machine involved?” but “was this made to help a reader, or to game a ranking?”

That purpose test is enforced by the same systems described in the spam sections. SpamBrain is one of “a variety of systems… that analyze patterns and signals to help us identify spam content, however it is produced”. And scaled content abuse — producing many pages for the primary purpose of manipulating rankings and not helping users — covers AI, human, or mixed production alike. A thousand pages written by hand for the sole purpose of ranking are scaled content abuse just as surely as a thousand pages generated by a model. The production method never was the violation; the scale and the purpose are.

Google’s guidance also answers the practical questions that follow. Appropriate use of AI or automation is not against its guidelines. Google will not ban AI content, because automation has long been used in publishing to create useful content. On misinformation, its systems look to surface high-quality information from reliable sources, and not information that contradicts well-established consensus on important topics — with an even greater emphasis on signals of reliability for health, civic, or financial information. On whether to use AI at all, Google’s test is a question back to the creator: “If you see AI as an essential way to help you produce content that is helpful and original, it might be useful to consider. If you see AI as an inexpensive, easy way to game search engine rankings, then no.”

Two rules about presentation complete the picture. Bylines are recommended where readers would expect them — the Who in the Who/How/Why test. AI or automation disclosures “are useful for content where someone might think ‘How was this created?'” — disclose where a reader would reasonably ask, not everywhere and not nowhere. And “giving AI an author byline is probably not the best way” to disclose AI involvement: an author byline implies a person, and a model is not one. The honest disclosure names the tool or the process; it does not invent a person.

The AI question connects directly to the features readers actually see in search. AI Overviews and AI Mode run on the same core ranking systems this article explains, and our guide to using Google’s AI search features without losing the classic web results covers that side of the experience. For site owners, the companion question is how content gets cited by AI answers at all — our GEO explainer on getting content cited by ChatGPT, Perplexity and AI Overviews covers the citation-earning work that sits on top of the ranking systems described here. The ranking systems decide what is good; the AI features decide what to show of it.

The Timeline: Every Confirmed Update from March 2024 to Today

Every date in this table comes from Google’s own Search Status Dashboard, read from its public incidents feed, except the March 2024 start date, which comes from Google’s own blog post. The dashboard is the only authoritative source for update dates. Feed timestamps are in UTC; the table gives dates only. No rows have been added for updates that were not re-verified from the dashboard in this run.

A note on the March 2024 row: the start date, 5 March 2024, is verified from Google’s own blog post. The completion date was not re-verified from the dashboard feed for this article — the feed returned only the ten most recent entries — so the finish cell is left as “April 2024 (verify on dashboard)” rather than stating a date that was not checked. No completion figure is given here, and none should be read into the table.

How should a reader read this table? First, notice how much rollout lengths vary: from under a day for the March 2026 spam update to about three weeks for the December 2025 core update. Google publishes no standard length, and none should be assumed — a rollout that is still running after a week is not necessarily a large one, and a rollout that finished in two days was not necessarily a small one. Second, notice the pattern of 2026: three core updates (March, May, November) and four spam updates (March, June, August, September), with the September spam update running longer than the other spam updates of the year. Third, notice what is different about two rows. The February 2026 update is a Discover update, not a Search update: it was designed to improve the quality of Discover overall, it reached English-language users in the US first, and it was to expand to all countries and languages later. And the March 2024 core update stands apart from every other row: Google described it as more complex than its usual core updates, involving changes to multiple core systems, with a rollout of up to a month and more ranking fluctuation than a regular core update — and it arrived on the same day as a spam update and three new spam policies.

Finally, notice what the table does not contain. Google does not publish impact numbers for any of these updates — not a share of queries affected, not a count of sites hit, not a traffic-loss range. The effects vary by site and by query, and any figure claiming to measure an update’s impact comes from a third party’s sample, not from Google. If you see a percentage attached to an update, ask who measured it and how. Google did not.

How to Tell If an Update Hit Your Site

The most common mistake in this whole subject is also the easiest to make: seeing traffic fall, finding the nearest update on a calendar, and blaming it. Google’s own diagnosis method exists to prevent exactly that error, and it starts with a discipline most people skip — waiting. The method uses Search Console, and it has four steps.

Step one: confirm the rollout has finished. Check the Search Status Dashboard. If the update is still rolling out — as the November 2026 core update is at the time of writing — you cannot yet diagnose anything, because rankings are still moving. Analysing a drop mid-rollout is analysing a moving target.

Step two: wait at least a full week after the rollout completes. Google’s instruction is explicit: wait at least a full week after completion before analysing Search Console. A week of settled data is the minimum for a reading that means anything. A single day’s drop, or a weekend’s dip, is noise until a full week says otherwise.

Step three: compare the right weeks. Compare the week after the rollout completed with the week before the rollout started. Not this week against last week in general, and not a day against the day before — the week after completion against the week before the rollout began. That comparison isolates the update’s window from everything else that moves traffic.

Step four: look at what actually moved. Review your top pages and queries, and analyse the search types separately — Web, Images, Video, and News — because an update can affect one search type and not the others, and a blended total can hide that. Google gives two worked examples for judging scale: a small drop, from position 2 to position 4, needs no drastic action. A large drop, from position 4 to position 29, needs a deeper assessment. The numbers are positions, not traffic, and the point is proportion: a small slide is a reassessment; a cliff is a signal that something on the page, or about the page, needs a closer look.

Then there is the mistake the method is designed to catch: blaming the nearest update for a drop that started before it. If your traffic began falling before the rollout started, the update did not cause it — something else did, and the update is a coincidence. Check the dashboard dates first, then check when your own decline began. If the dates do not line up, look elsewhere: a site change, a seasonal pattern, a lost link, a competitor’s improvement, or one of the smaller unannounced core updates that Google never lists. The dashboard tells you when the named updates ran. It does not tell you that a named update caused your drop. Only your own data, compared across the right weeks, can do that.

Recovery: What Google Actually Tells You to Do

Recovery is where the advice around Google updates is at its worst, and Google’s own guidance is at its most honest. The first thing it says is what not to do: there are no quick fixes. Do not make hasty changes in the hope of reversing a drop. A page that fell after a core update was reassessed, not penalised, and the response is to make meaningful, sustainable improvements for users — not to chase the update.

The working tool is the self-assessment from the helpful content section of this guide. Run the questions honestly on the pages that fell: is the information original, substantial, and complete? Does it offer analysis beyond the obvious? Does it avoid simply copying or rewriting sources? Does it demonstrate first-hand expertise and depth of knowledge? Would a reader leave having learned enough to achieve their goal? Google suggests doing this assessment with people unaffiliated with your site, if possible — fresh eyes catch self-deception that the author cannot. The questions are the diagnosis; the answers are the work list.

Deleting content is a last resort. Google says so directly, and adds a pointed observation: deleting whole sections of a site is “likely a sign those sections were created for search engines first, and not people”. If a section only existed to attract search traffic, removing it is an admission about why it was built, not a recovery strategy. Remove pages that are genuinely unhelpful, outdated, or misleading — but do not delete pages that dropped simply because they dropped. A dropped page that is still useful to readers is a page to improve, not a page to erase.

On timing, Google gives the only two statements it has ever given, and both are ranges, not promises. Improvements “can take effect in a few days, but it could take several months”. And you do not have to wait for the next major core update: smaller core updates run continuously and are not announced, so improvements can show effect without waiting for the next big one. What “recovery” realistically looks like, then, is gradual and partial and never guaranteed. Some pages come back. Some come back part of the way. Some do not come back at all, because the field around them improved faster. Anyone who promises a recovery date, or a recovery at the next core update, is promising something Google itself does not promise.

The spam path is different, and it should not be confused with the core path. If a site violated a spam policy, the fix is to stop the violation and clean up what was done — remove the cloaking, the doorway pages, the hidden text, the spammy links, the scraped content — and then wait for the automated systems to learn over a period of months that the site complies. The timeline is months by design. And the link-spam exception applies here in full: once Google removes the effect of spammy links, the ranking benefit they gave is lost and cannot be regained by fixing the links. A site that bought its rankings with links does not get them back by disavowing them; it has to earn rankings another way.

For the question of what SEO work actually helps after a drop, as opposed to the “hacks” Google itself dismisses, our AEO vs GEO vs SEO guide separates genuine SEO work from sold shortcuts. The recovery work itself, though, is the same work this guide has described throughout: original, expert, people-first content, a clean spam record, and patience measured in months rather than days.

Myths and Mistakes

“Core updates are penalties.” They are not. A penalty is enforcement against a violation; a core update is a broad reassessment of how content measures up. Pages that drop after a core update have not necessarily done anything wrong — other pages may simply have been reassessed as more helpful. Treating a core update as a penalty leads to the wrong response: hunting for a violation that does not exist instead of improving the content.

“There is a fixed schedule.” There is not. Google says core updates happen “several times a year”. No calendar is published, no dates are announced in advance, and no fixed annual count exists. Anyone selling a calendar of coming core updates is selling a guess.

“Recovery happens at the next core update.” It does not have to. Smaller core updates run continuously and are not announced, so improvements to your content can show effect without waiting for the next major core update. Waiting for a named update before improving a page is waiting for a date that has nothing to do with the work.

“A third-party authority score tells you how Google sees your site.” It does not. Google says third-party “reputation” or “authority” scores “don’t correspond to any of Google’s own signals nor come from Google”. A score sold by a tool is that tool’s opinion, measured its own way. It is not a window into Google’s systems.

“Delete the pages that dropped.” Deleting content is a last resort, not a first response. Google’s own observation is that deleting whole sections is likely a sign those sections were created for search engines first and not people. Improve the pages that are still useful; remove only what is genuinely unhelpful.

“AI content is banned.” It is not. Google rewards quality however content is produced, and automation has long been used to produce useful content such as sports scores, weather forecasts, and transcripts. What violates the spam policies is using automation — including AI — with the primary purpose of manipulating rankings. Purpose is the test, not production method.

“I can guarantee a number one ranking.” No one can. Google’s own SEO guidance says there are no secrets that will automatically rank a site first in Google. A guarantee of a top ranking is a promise about a system the person making it does not control, and it should be treated as a warning sign about whoever is making it.

“The update that happened last week must be why my traffic fell.” Check the dashboard dates first. If your drop started before the rollout began, the update did not cause it. If the rollout is still running, you cannot yet diagnose anything. And if no update was running at all, the cause is elsewhere — a site change, a seasonal pattern, or one of the smaller unannounced core updates that Google never lists. The nearest update is the easiest explanation, and very often the wrong one.

Your Checklist: What to Do This Week

Here is the work, in order, for a site owner reading this guide in the week of 11 November 2026.

1. Check the dashboard. Is anything rolling out right now? Yes: the November 2026 core update began on 10 November 2026 and was still rolling out at the check for this article. Do not diagnose anything until it finishes. Bookmark the Search Status Dashboard and check it before you check anything else.

2. Check Search Console the right way. Wait at least a full week after the November update completes, then compare that week with the week before the rollout started. Review your top pages and queries, and analyse Web, Images, Video, and News separately. A small slide in position needs no drastic action; a large drop needs a deeper assessment.

3. Run the self-assessment on your most-affected pages. Use the people-first questions from the helpful content section: original information, substantial and complete coverage, analysis beyond the obvious, clear sourcing and expertise, a reader who leaves having learned enough to achieve their goal. Ask someone unaffiliated with your site to run it too, if you can.

4. Fix what is genuinely unhelpful, in priority order. Start with the pages that matter most — the ones with the most traffic, the most links, or the most importance to your readers — and make meaningful, sustainable improvements for users. Do not make hasty changes, and do not chase the update.

5. Check spam-policy compliance if a spam update overlaps your dates. The September 2026 spam update ran from 24 September to 8 October 2026. If your drop falls in that window, the question is not whether your content is helpful enough but whether it violates a policy — and the fix is to stop the violation and wait months for the systems to learn.

6. Do not touch what is performing well. A page that held its position through an update is a page the reassessment judged well. Leave it alone. The most expensive mistake after an update is “fixing” pages that were never broken.

7. Re-check the dashboard for the November update’s completion and note the date. When Google posts the completion message, write the date down. It is the start of your one-week waiting period, and the anchor for the before-and-after comparison that follows. Until then, the honest status is the one this guide opened with: a core update is rolling out, Google has said nothing about what it changes, and the work is the same work Google has described since 2022 — helpful, people-first content. No update changes that.

Sources

All facts in this guide were checked against Google’s own documentation, blog, and status feed on 11 November 2026. The November 2026 core update status was re-checked directly against the Search Status Dashboard incidents feed (https://status.search.google.com/incidents.json) on 11 November 2026: the update began on 10 November 2026 and was still rolling out, with no completion message posted. Google has published nothing about what the November 2026 core update changes, and this guide speculates about nothing on that point. Dates in the article are given as dates only; feed timestamps are in UTC.

• Google Search Status Dashboard (incidents feed: https://status.search.google.com/incidents.json) — every date in the timeline table; November 2026 core update status re-checked 11 November 2026
• Google Search Central, “Google Search’s core updates and your website” — core update frequency and scope, the restaurant-list analogy, the Search Console diagnosis method, quick fixes, deleting content, recovery timing, smaller unannounced core updates; checked 11 November 2026
• Google Search Central, “A guide to Google Search ranking systems” — the named ranking systems and their plain-language descriptions, page-level ranking with site-wide signals, retired systems; checked 11 November 2026
• Google Search Central, “Creating helpful, reliable, people-first content” — the self-assessment questions, E-E-A-T, people-first tests, search-engine-first warning signs, Who/How/Why; checked 11 November 2026
• Google Search Central, “Google Search spam updates and your site” — spam updates as notable improvements, SpamBrain, consequences for violators, the link-spam exception; checked 11 November 2026
• Google Search Central, “Spam Policies for Google Web Search” — all sixteen spam policies and the demotion and removal practices; checked 11 November 2026
• Google Search Central, “Google Search’s reviews system and your website” — what the reviews system rewards and evaluates, page-level evaluation, the eleven languages, ongoing improvement, recovery; checked 11 November 2026
• Google Search Central, “How To Write Reviews” — best practices for writing reviews; checked 11 November 2026
• Google Search Central Blog, “What web creators should know about our March 2024 core update and new spam policies” (5 Mar 2024) — the March 2024 core update, the helpful content system folding into core, the three new spam policies, site reputation abuse, third-party scores; checked 11 November 2026
• Google Search Central Blog, “Google Search’s guidance about AI-generated content” (8 Feb 2023) — quality over production method, automation as spam only when its purpose is manipulating rankings, disclosures, author bylines; checked 11 November 2026