GSC Blog: Core Updates and the Evolution of Algorithm Iterations (2010–2026)

A thematic distillation of 23 official Google Search Central blog announcements.Not ordered chronologically; organized around an "algorithmic mental model" —— if you want to understand how Google itself explains algorithm evolution, this is less effort than reading a timeline.

Of the 23 posts, 2 are non-English stubs (2010-06 Caffeine in German / 2019-12 Events in Spanish/German). Their content cannot be directly quoted, but metadata from the stubs themselves can be used in arranging the timeline.

Part 1 — Google’s Official Mental Model of Core Updates

Google repeatedly uses the same passage to define broad core updates, almost as a reusable template.Key takeaway: Core updates are not a penalty mechanism, not a spam action, and do not target specific pages or sites —— they are a process of "reassessing the relative value of content across the entire web.".

1.1 Standard definition (first core-updates announcement in 2019 + reused in every subsequent announcement)

Source:2019/08/core-updates.en.md —— this post is the "bible" for all later core update announcements; every subsequent announcement cites it.

Several times a year, we make significant, broad changes to our search algorithms and systems. We refer to these as "core updates." They're designed to ensure that overall, we're delivering on our mission to present relevant and authoritative content to searchers.

We release broad core updates several times a year, with the goal of ensuring that overall, we are presenting users with relevant, authoritative content.

1.2 Anti-penalty narrative (Google’s repeated emphasis: "don’t make the wrong changes")

There's nothing wrong with pages that may perform less well in a core update. They haven't violated our webmaster guidelines nor been subjected to a manual or algorithmic action... In fact, there's nothing in a core update that targets specific pages or sites. Instead, the changes are about improving how our systems assess content overall. These changes may cause some pages that were previously under-rewarded to do better.

—— 2019/08/core-updates,2022/05/may-2022-core-update Verbatim quote of this passage

Why: This passage is reused almost word-for-word with every core update (present in 2019, 2022/05, 2024/03, and 2024/08), with the goal of discouraging site owners from "fixing the wrong things." Google’s subtext is: your drop in rankings is not because you did something wrong, but because others did things right —— the algorithm simply reassessed relative value.

1.3 The movie list analogy (Google’s favorite illustrative story)

One way to think of how a core update operates is to imagine you made a list of the top 100 movies in 2015. A few years later in 2019, you refresh the list. It's going to naturally change. Some new and wonderful movies that never existed before will now be candidates for inclusion. You might also reassess some films and realize they deserved a higher place on the list than they had before.

—— 2019/08/core-updates

This is Google’s most classic metaphor for explaining core updates:A change in the rankings list ≠ old movies got worse; rather, new movies are better + the value of older movies is being re-recognized..

1.4 Why it exists: countering the two-way drift of "content degradation + rising user expectations"

On the web page side: from static HTML → rich media → real-time streams (the core driving force in the Caffeine 2010 era);
On the user side:"Searchers want to find the latest relevant content and publishers expect to be found the instant they publish."(2010/06/caffeine);
On the algorithm side: needs to continuously re-weight 200+ signals (the 2016 Penguin announcement reused this number);

So a core update is not "fixing a bug," but "continuously calibrating the relative weights of 200+ signals."

Part 2 — Timeline: From Caffeine to 2026 Discover

2.1 Division into three major eras

Era	Core narrative	Representative updates
2010 Indexing Era	Changing from "batch indexing" to "continuous incremental indexing"	Caffeine (2010/06)
2012–2016 Algorithm Wave	Single-point signals (Panda/Penguin) became standalone algorithms → later absorbed into core	Penguin → core (2016/09)
2019–2022 Institutionalization of Core Updates	Naming of broad core updates standardized, released roughly quarterly	2019/08 first guidance, 2022/05
2022–2024 Helpful Content Era	Added a site-wide signal to counter "content written for search engines"; eventually integrated into core	Helpful Content Update 2022/08 → absorbed in 2024/03
2024–2026 Convergence and Simplification	Core updates increasingly become "complex multi-system fusions"; product features begin to be trimmed	March 2024 core (multi-system), 2025/11 SERP simplification, 2026/02 Discover core

2.2 Key milestones table

Date	Announcement	Nature	Core mechanism
2010-06-01	Caffeine/Mayday (German stub)	Indexing architecture	Long-tail changes
2010-06-09	Caffeine new indexing system officially announced	Indexing architecture	From layered batch indexing → continuous incremental indexing, 50% fresher; 100M GB of storage
2016-09-23	Penguin is now part of our core	Algorithm integration	Penguin moved from standalone refresh → real-time and granular,no longer announced as a separate refresh; merged into core ranking
2019-08-01	What site owners should know about core updates	Institutionalization	First systematic explanation of broad core updates, introduced the E-A-T self-assessment checklist; became the master template for all subsequent core update announcements
2019-12-01	Events launched in Germany/Spain (non-English stub)	Feature launch	Unrelated to core updates, part of SERP feature expansion
2020-05-28	Evaluating page experience for a better web	Signal introduction preview	Previewed that Core Web Vitals (LCP/FID/CLS) plus existing page experience signals would be combined into a single "page experience" ranking signal, delayed to 2021;removed the AMP requirement for Top Stories
2020-11-10	Timing for page experience	Timing confirmation	Locked in launch for 2021/05; Lighthouse/PageSpeed usage rose 70%
2021-04-08	April 2021 Product Reviews Update	First topic-specific update	English-only, targeting "thin content reviews"; introduced an in-depth/expert/originality self-assessment checklist
2021-04-19	More details on page experience update	Timing refinement	Gradual rollout starting mid-2021/06, fully rolled out by 2021/08; AMP requirement removed from Top Stories
2021-07-26	Link spam update + qualifying links reminder	Spam update	Reminded about rel="sponsored"; launched a new link spam algorithm
2021-08-04	Simplifying the Page Experience report	Report/clarification	Safe Browsing is not a ranking signal (key clarification); Ad Experience separated from the Page Experience report
2021-11-04	Bringing page experience to desktop	Scope expansion	Starting 2022/02, desktop also uses page experience; mobile-friendliness applies only to mobile
2021-12-01	December 2021 Product Reviews Update	Topic iteration	Added two best practices: real-world evidence (visuals/audio) + multi-seller links
2022-03-23	Improving Product Review ranking, one year on	Topic iteration	Added FAQ clarifying that "ranked lists/comparison reviews also count"; added first-hand evidence requirement for "best of" recommendations
2022-05-25	May 2022 core update	Regular core update	1–2 week rollout; verbatim reuse of the 2019 template language
2022-08-18	Helpful Content Update(HCU)	New mechanism	Added a new site-wide signal (not a spam action), ML classifier runs continuously; the "people-first content" concept officially debuted
2022-12-14	December 2022 Link Spam Update	Spam update	First mention of SpamBrain AI detecting link spam; identifying sites that buy/sell links
2023-04-19	The role of page experience in creating helpful content	Major pivot	Page experience retreated from being a "signal/system" to being a "guiding principle"; acknowledged that "there is no single page experience signal"; Page Experience report was split up
2023-10-04	October 2023 Spam Update	Spam update	Multilingual coverage (Turkish/Vietnamese/Indonesian/Hindi/Chinese, etc.); targeted cloaking / hacked / auto-generated / scraped content
2024-03-05	March 2024 core update + new spam policies	Major convergence	"more complex than usual, involving multiple core systems"; HCU was absorbed back into core; three new spam policies:expired domain abuse / scaled content abuse / site reputation abuse; the latter took effect on 2024/05/05
2024-08-15	August 2024 core update	Corrective core update	Explicitly stated that they "heard creators’ feedback," more rewards for small independent sites; captures site improvements
2025-11-05	Search results page simplification update	Product feature trimming	Not a ranking update —— cut infrequently used features on the SERP; starting 2026/01, the Search Console API will remove the corresponding structured data types
2026-02-05	February 2026 Discover Core Update	Discover-specific core update	A broad core update specifically for Discover;reduce sensational/clickbait;topic-by-topic expertise(different topics on the same site are evaluated separately)

Part 3 — Topic-Specific Updates: How the Spam / Product Reviews / Helpful Content Lines Converged

3.1 The Spam Update line (from rule-driven → AI-driven)

Penguin being merged into core in 2016 was the watershed. The Penguin announcement (2016/09) declared two things:

Penguin is now real-time... we're not going to comment on future refreshes. Penguin is now more granular. Penguin now devalues spam by adjusting ranking based on spam signals, rather than affecting ranking of the whole site.

Why: from "penalizing entire sites + periodic announced refreshes" → "signal-based, real-time, page-level."

Key evolutions in spam announcements after that:

2021/07 Link Spam Update: still rule-based + algorithmic;
2022/12 December Link Spam Update:First explicit disclosure that SpamBrain (AI-based spam-prevention) was used for link spam;
2023/10 October Spam Update: multilingual coverage, community-driven (spam types reported by the community drove updates in reverse);
Major 2024/03 spam policy changes: from "anti auto-generated content" → "anti scaled content abuse" (regardless of whether produced by humans, machines, or a mix, as long as it is scaled + manipulates rankings it counts).

Our long-standing spam policy has been that use of automation, including generative AI, is spam if the primary purpose is manipulating ranking in Search results. The updated policy is in the same spirit of our previous policy and based on the same principle. It's been expanded to account for more sophisticated scaled content creation methods where it isn't always clear whether low quality content was created purely through automation.

—— 2024/03/core-update-spam-policies

Why: This is Google’s official stance after the explosion of AI-generated content —— it does not oppose AI, it opposes scale + manipulative intent.

3.2 The Product Reviews topic line (standalone → merged back into core)

First launch 2021/04: English-only, countering "thin reviews"; introduced the "in-depth/expert" checklist;
Second update 2021/12: added "real-world evidence" + "multi-seller links";
Third update 2022/03: covered ranked lists / "best of" recommendations;
Aftermath: The Product Reviews Update was gradually merged into core updates starting in 2023; after 2024/03, official announcements basically no longer separately mentioned a "product reviews update."

Why build a separate line then merge it back into core: Google experimentally made "product reviews" a standalone signal, and after validating that it worked,it deposited the methodology (in-depth / first-hand / real-world testing) into the helpful content evaluation system within core ranking.

3.3 The Helpful Content Update (HCU) line —— the most dramatic merge back into core

The first HCU release in 2022/08 was a new mechanism:

Not a spam action, not a manual action;
it was site-wide ML classifier signal(key quote:This classifier process is entirely automated, using a machine-learning model);
runs continuously, can monitor new and old sites;
"Removing unhelpful content can help your other content rank" —— the first explicit statement that low-quality content can drag down an entire site.

2024/03 announcement: HCU absorbed back into core:

The March 2024 core update is a more complex update than our usual core updates, involving changes to multiple core systems. It also marks an evolution in how we identify the helpfulness of content... There's no longer one signal or system used to do this.

Why this is a big deal:Google explicitly told the SEO community —stop researching what the "helpful content classifier" is and how to pass it — it has already been broken up and distributed across multiple core systems. This turned HCU from an object that could be individually "diagnosed/evaded" into a "systemic judgment".

Part 4 — The position of Page Experience / Core Web Vitals: from "ranking signal" → "tie-breaker" → "guidance"

This is the most dramatic storyline across the 23 posts, worth tracking on its own:

4.1 Rise period (2020–2022): Grand introduction

2020/05 evaluating-page-experience:Teased that CWV (LCP/FID/CLS) + mobile-friendly + HTTPS + intrusive interstitials would be combined into "a new signal"; key quote:a holistic picture of the quality of a user's experience on a web page;
2020/11 timing:Locked in for 2021/05;
2021/04 more details:Changed to a gradual rollout from mid-2021/06 to the end of 2021/08;teased SXG / Top Stories removing AMP;
2021/08 simplifying:Key clarification that Safe Browsing is not a ranking signal(i.e., the PE signal does not include Safe Browsing);
2021/11 desktop:Extended to desktop;

But even during this grand rollout period, Google had already planted a hedging narrative:

While all of the components of page experience are important, we will prioritize pages with the best information overall, even if some aspects of page experience are subpar. A good page experience doesn't override having great, relevant content. However, in cases where there are multiple pages that have similar content, page experience becomes much more important for visibility in Search.

—— 2020/05/evaluating-page-experience (this passage appeared repeatedly in multiple posts from 2020–2021)

Why:From the very beginning Google described PE as a tie-breaker (only having a noticeable effect when content is of equal quality), not a primary ranking factor.

4.2 Turning point (2023/04): Quietly "walking back"

2023/04/page-experience-in-search.en.md is the key turning point of the entire PE story. On the surface it was "integrating PE into helpful content guidance", but in substance Google was walking back:

There is no single signal. Our core ranking systems look at a variety of signals that align with overall page experience.

While not all of these may be directly used to inform ranking, we do find that all of these aspects of page experience align with success in search ranking, and are worth attention.

The page experience update was a concept to describe a set of key page experience aspects for site owners to focus on... It was not a separate ranking system, and it did not combine all these signals into one single "page experience" signal.

Why this is a walk-back:The messaging in 2020–2021 all said "a new signal that combines...", but in 2023 Google directly rewrote history — "it was a concept, not a signal".

At the same time, the following were removed:

Starting 2023/12/01, the Mobile Usability report, Mobile-Friendly Test tool, and API were all retired;
Starting 2023/11/08, the "Good page experience" search appearance filter was also retired;
The Page Experience report was changed into "a dashboard that links to guidance".

4.3 INP replaces FID (2024/03/12)

The only technical PE update: Interaction to Next Paint replaced First Input Delay in CWV.

4.4 Current position (2023+): Background guidance

PE's current official positioning:

Not a single signal;
It is holistic guidance, sharing the same origin as helpful content;
It is a tie-breaker;
CWV is still "highly recommended" but"good stats don't guarantee good rankings".

Part 5 — Core beliefs Google repeatedly emphasized (golden lines that appeared 3+ times across the 23 posts)

5.1 "Quality comes before optimization"

People-first content creators focus first on creating satisfying content, while also utilizing SEO best practices to bring searchers additional value.

—— 2022/08 HCU (cited in multiple subsequent posts)

5.2 "There is nothing specific to fix"

Repeated in every core update announcement:"There might not be anything to fix at all."

5.3 "Making changes does not guarantee recovery"

Do keep in mind that improvements made by site owners aren't a guarantee of recovery, nor do pages have any static or guaranteed position in our search results. If there's more deserving content, that will continue to rank well with our systems.

—— 2019/08

5.4 "Google does not understand content the way humans do; it only looks at signals"

It's also important to understand that search engines like Google do not understand content the way human beings do. Instead, we look for signals we can gather about content and understand how those correlate with how humans assess relevance.

—— 2019/08

5.5 "Raters do not directly control rankings; they only calibrate the algorithm"

Rater data is not used directly in our ranking algorithms. Rather, we use them as a restaurant might get feedback cards from diners.

—— 2019/08

5.6 "There are multiple signals, and the specific ones are not disclosed"

We use many more [signals], which we don't disclose to help protect the integrity of our results.

—— 2019/08

5.7 "Topic-by-topic expertise" (newly added in 2026)

Our systems are designed to identify expertise on a topic-by-topic basis. So whether a site has expertise in multiple areas or has a deep focus on a single topic, there's equal opportunity to show up in Discover.

—— 2026/02 Discover core (using a local news site + its gardening section as an example)

This is a mirror of the "site reputation abuse" policy — since Google can identify expertise across different subtopics, it can also identify third-party content that a host parasitically attaches using its own reputation.

Part 6 — Reverse guidance: Things Google explicitly says it "will not do"

Boundaries Google explicitly drew with negative statements across the 23 announcements:

6.1 Will not publish the specific list of 200+ signals

our algorithms rely on more than 200 unique signals or "clues"... (2016 Penguin) We use many more, which we don't disclose to help protect the integrity of our results. (2019)

6.2 Will no longer announce every spam algorithm refresh

we're not going to comment on future refreshes. (when Penguin went real-time in 2016)

6.3 There is no "single page experience signal"

There is no single signal. (2023/04)

6.4 Does not accept "3rd party site authority scores"

some third-party services provide "reputation" or "authority" scores for sites. These don't correspond to any of Google's own signals nor come from Google. (2024/03)

(This is an official rejection of metrics such as Moz DA / Ahrefs DR)

6.5 There is no "preferred word count"

Are you writing to a particular word count because you've heard or read that Google has a preferred word count? (No, we don't). (2022/08 HCU)

6.6 E-A-T is not a ranking factor

Since we originally wrote this post, we have been occasionally asked if E-A-T is a ranking factor. Our automated systems use a mix of many different signals to rank great content. We've tried to make this mix align what human beings would agree is great content as they would assess it according to E-A-T criteria. (2020/03 added to 2019/08 post)

Why:E-A-T is a rater evaluation framework, not an algorithmic signal; the algorithm aligns with human judgments made against E-A-T standards, and does not implement E-A-T.

6.7 Improvement does not equal guaranteed recovery

See 5.3. Google refuses to give a promise that "if you do the right thing you will definitely come back".

6.8 Safe Browsing / Ad Experience are not ranking signals

we're clarifying that Safe Browsing isn't used as a ranking signal (2021/08) the Ad Experience report was never used as a factor for page experience (2021/08)

6.9 AI content is not "automatically spam"

Our long-standing spam policy has been that use of automation, including generative AI, is spam if the primary purpose is manipulating ranking. (2024/03)

"If" is the key word — using AI to write people-first content does not violate the rules; using any means (including purely manual work) to produce content at scale for volume does violate the rules.

Part 7 — Cross-cuts: Hidden narratives outside the 23 posts

Several narratives that Google did not state but that emerged after reading all 23 posts:

7.1 "Merging back into core" is Google's reclamation mechanism

Penguin (standalone in 2012 → merged back in 2016), Product Reviews (standalone in 2021 → quietly merged back in 2024), HCU (standalone in 2022 → explicitly merged back in 2024).Google's methodology:Use standalone systems for rapid experimentation + user feedback, and once validated as effective, absorb them into core so site owners cannot optimize for a single point.

7.2 Announcement density reflects public opinion pressure from the SEO community

After 2022/08 HCU → continuous announcements + FAQ + 2024/03 complex core + 2024/08 "we heard creators' feedback" → this indicates HCU triggered large-scale backlash from small publishers, and Google has been continuously doing public relations.

7.3 The "expansion → simplification" cycle on the timeline

2020–2022 was the PE signal expansion period;
2023–2026 is the simplification period (retiring Mobile Usability / Mobile-Friendly Test / Page Experience report / various SERP features / cutting structured data types from the Search Console API starting 2026/01).

Why:Google periodically collapses "many signals + many reports" back into "fewer signals + fewer reports + more holistic guidance", first to reduce site owners' room to "optimize for appearances", and second to lower its own product maintenance costs.

7.4 Discover is a new standalone battlefield (2026/02)

The first time a broad core update was released specifically for Discover — migrating methodology validated on Search (reduce clickbait + topic expertise) to Discover. This foreshadows that Discover will become the testing ground for the next wave of ranking evolution.

Appendix — List of the 23 original documents

Sorted chronologically, marked as stub (non-English shell) and substantive (with content):

#	Date	slug	Type
1	2010-06-01	caffeine-oder-mayday	stub (German)
2	2010-06-09	our-new-search-index-caffeine	substantive
3	2016-09-23	penguin-is-now-part-of-our-core	substantive
4	2019-08-01	core-updates	substantive (master version)
5	2019-12-01	events-launching-germany-spain	stub (Spanish/German)
6	2020-05-28	evaluating-page-experience	substantive
7	2020-11-10	timing-for-page-experience	substantive
8	2021-04-08	product-reviews-update	substantive
9	2021-04-19	more-details-page-experience	substantive
10	2021-07-26	link-tagging-and-link-spam-update	substantive
11	2021-08-04	simplifying-the-page-experience-report	substantive
12	2021-11-04	bringing-page-experience-to-desktop	substantive
13	2021-12-01	product-reviews-update-and-your-site	substantive
14	2022-03-23	product-review-ranking-one-year-on	substantive
15	2022-05-25	may-2022-core-update	substantive
16	2022-08-18	helpful-content-update	substantive
17	2022-12-14	december-22-link-spam-update	substantive
18	2023-04-19	page-experience-in-search	substantive (walk-back turning point)
19	2023-10-04	october-2023-spam-update	substantive
20	2024-03-05	core-update-spam-policies	substantive (major integration)
21	2024-08-15	august-2024-core-update	substantive
22	2025-11-05	update-on-our-efforts	substantive (SERP simplification)
23	2026-02-05	discover-core-update	substantive (Discover debut)