Google Algorithm Updates: A Timeline
Last updated
Google makes thousands of algorithm changes per year. Most are minor, unannounced, and imperceptible. The ones listed here were large enough to cause measurable ranking shifts, were officially confirmed, or introduced a fundamentally new ranking signal.
Dates and durations for confirmed updates from 2023 onwards are taken from Google’s own Search Status Dashboard,1 which is the authoritative record and frequently disagrees with the dates circulated in SEO coverage. Current to 21 August 2026: the most recent confirmed update is the August 2026 spam update, which completed on 21 August.
2011
Panda (February 2011)
Target: Low-quality, thin, and duplicate content.
Panda assigned a site-wide quality score based on content quality. Sites with a high proportion of thin, scraped, or auto-generated content saw sitewide ranking drops rather than page-level ones. Content farms, article directories, and sites with large amounts of boilerplate text were hit hardest.
Panda was initially a filter applied periodically (every few weeks). It became part of Google’s core algorithm and began rolling out continuously in 2016.
Signal: Content must have genuine depth and utility. Thin pages drag the whole site down.
Freshness Update (November 2011)
Target: Query types that benefit from recent content.
Increased the weight of freshness as a ranking signal for queries where users expect current information: news, trending topics, recurring events. Not relevant for evergreen queries.
2012
Venice (February 2012)
Target: Local search results.
Made local results more prominent for queries with implied local intent (e.g., “plumber” without a city name), using the searcher’s location to determine relevance.
Penguin (April 2012)
Target: Manipulative link building: link schemes, paid links, low-quality directory links, exact-match anchor text over-optimisation.
Penguin penalised sites with link profiles that showed signs of manipulation. Unlike Panda (content quality), Penguin was link-focused. Sites hit by Penguin needed to remove or disavow bad links before recovery.
Penguin updates were initially infrequent (several months apart). In September 2016, Penguin became part of the core algorithm and began updating in real time. Penalties and recoveries now happen as Google recrawls and reassesses links.
Signal: Links must be earned, not built. Anchor text diversity matters.
Exact Match Domain (EMD) Update (September 2012)
Target: Low-quality sites ranking purely because their domain name matched a search query.
Reduced the ranking boost from exact-match keyword domains (e.g., best-seo-tools.com) when the site had little quality to back it up. High-quality sites with keyword domains were not affected.
Page Layout (January 2012, updated October 2012)
Target: Pages with excessive above-the-fold advertising that pushed content below the fold.
2013
Hummingbird (August 2013, announced September 2013)
Target: Conversational queries and semantic search.
Hummingbird was a near-complete rewrite of Google’s core search algorithm, not a penalty or filter. It enabled Google to understand the intent behind a query rather than matching individual keywords. Introduced better handling of long-tail, conversational, and question-based queries.
Signal: Write for topics and intent, not individual keyword strings.
Payday Loan (June 2013)
Target: Highly spammy queries: payday loans, pharmaceuticals, gambling. These were areas where results were dominated by manipulative tactics.
2014
Pigeon (July 2014)
Target: Local search quality and distance signals.
Significantly changed how Google ranked local results, tying local ranking more closely to core web ranking signals. Businesses with stronger overall web presence and proximity to the searcher saw changes.
HTTPS as a ranking signal (August 2014)
Google confirmed HTTPS as a lightweight ranking signal, encouraging the adoption of SSL. Now a baseline requirement. HTTP sites are flagged as insecure in Chrome and are at a disadvantage.
Panda 4.0 (May 2014)
A major Panda refresh that affected a larger proportion of queries than previous updates. eBay and Ask.com were widely cited as large-scale casualties.
2015
Mobilegeddon / Mobile-Friendly Update (April 2015)
Target: Mobile usability. Pages that failed Google’s mobile-friendly test were demoted in mobile search results.
The first major update to split rankings by device. Effectively made mobile usability a ranking requirement for mobile search. A second wave rolled out in May 2016.
RankBrain (announced October 2015, rolled out earlier)
Target: Query interpretation for ambiguous or never-seen-before queries.
RankBrain is a machine learning component that interprets queries Google has not seen before (a significant proportion of daily searches). It maps unfamiliar queries to known concepts to produce relevant results.
A Google search quality strategist described it as the third most important ranking signal in 2016, after content and links.2 Treat that as a statement of its moment rather than a current ranking: Google has since moved away from ranking its own signals that way, and Gary Illyes has said links are not a top-three factor and have not been for some time. RankBrain is now part of the core ranking infrastructure rather than a distinct “update.”
Signal: Optimise for topic relevance and user satisfaction, not exact keyword strings.
2016
Possum (September 2016)
Target: Local search, specifically filtering duplicate local listings and improving results for searchers near the business but outside city limits.
Made Google’s filter for near-duplicate local business listings more aggressive. Businesses with similar names or at similar addresses saw listings filtered from map packs.
Penguin 4.0 (September 2016)
Penguin became real-time and part of the core algorithm. Links Google considers spammy are now devalued rather than triggering a site-wide penalty. Recovery from Penguin no longer requires waiting for the next update cycle.
2017–2018
Fred (March 2017)
Unofficial name for a core update targeting sites prioritising revenue over user value. This included thin affiliate content, heavy ad-to-content ratios, and content existing primarily to generate ad impressions.
Broad Core Updates become regular (2018 onwards)
From 2018, Google began officially confirming “broad core algorithm updates” multiple times per year. These are updates to the core ranking systems that can affect any site or query type. Google’s guidance: sites that drop after a core update have not been penalised. Google’s assessment of what constitutes quality has shifted.
Medic Update (August 2018)
Unofficial name for a broad core update with disproportionate impact on health, finance, and legal sites. YMYL (Your Money or Your Life) categories saw low-quality information where it could harm users.
Widely attributed to increased weight on E-A-T signals (Expertise, Authoritativeness, Trustworthiness). Authors, credentials, and organisational reputation became more visible ranking considerations for YMYL content.
2019
BERT (October 2019)
Target: Natural language understanding, particularly for complex queries.
BERT (Bidirectional Encoder Representations from Transformers) enabled Google to better understand context in search queries: the relationship between words in a query rather than treating each word in isolation. Particularly effective for prepositions and nuance: “parking on a hill with no curb” had previously returned generic parking results. BERT understood the specific scenario.
Initially applied to one in ten English queries.3 Now applied more broadly and in multiple languages.
Signal: Write naturally. Context and clarity matter more than keyword density.
2020
Passage Ranking (announced October 2020, rolling out 2021)
Google can now rank individual passages from a page for queries that match a specific section, even if the page as a whole is not the most relevant for the broader topic. Affects approximately 7% of queries.4
Core Updates (January, May, December 2020)
Three confirmed broad core updates in 2020, with the May update producing some of the largest ranking volatility observed that year.
2021
Page Experience / Core Web Vitals (June–September 2021)
Google added Core Web Vitals (then LCP, FID, CLS) as ranking signals, combined with existing page experience signals (HTTPS, mobile-friendly, no intrusive interstitials). Rollout confirmed complete in September 2021.
A confirmed ranking factor but described as a tiebreaker: relevant content with poor page experience can still rank above fast content that is less relevant.
The metric set has since changed. Interaction to Next Paint (INP) replaced First Input Delay as the responsiveness metric in March 2024: Google announced that the Chrome team “decided to promote INP as the new Core Web Vitals metric for responsiveness, effective March 2024, replacing FID”.5 Advice written before that date measures a metric Google no longer uses.
Spam Updates (June and November 2021)
Targeted link spam and web spam more broadly. Worked in conjunction with SpamBrain, Google’s AI-based spam detection system.
MUM (Multitask Unified Model)
Google announced MUM, a multimodal AI model significantly more capable than BERT. MUM can process text, images, and video simultaneously. Used selectively for complex queries and in features like Shopping and Search.
2022
Product Reviews Update (March 2022, and subsequent updates)
Targeted product review content. Rewarded reviews demonstrating first-hand experience: original photos, specific performance data, comparisons with similar products. Generic affiliate reviews written without hands-on use were demoted.
Multiple refreshes followed throughout 2022 and 2023, eventually consolidated into core updates.
Helpful Content Update (August 2022)
Target: Content created primarily for search engines rather than humans.
Introduced a site-wide classifier that identifies sites with a high proportion of “unhelpful” content: content written to match search queries without providing genuine value. Sites classified as producing primarily unhelpful content saw sitewide ranking impacts across the whole domain, not limited to the unhelpful pages themselves.6
The signal became a component of Google’s core ranking system. Subsequent core updates incorporated HCU signals.
Signal: Write for people, not crawlers. Content that exists solely to rank is exactly what this targets.
Link Spam Update (December 2022)
Expanded SpamBrain’s ability to detect and neutralise purchased links and large-scale link schemes. Google claimed the update nullified “unnatural links at scale.”
2023
March 2023 Broad Core Update
Significant ranking volatility, particularly for health, finance, and news content. Ran concurrently with a Product Reviews Update.
August 2023 Broad Core Update
Notable for unusually large and prolonged volatility. Some sites that had recovered from the HCU saw further drops.
September 2023 Helpful Content Update
Ran 14–27 September 2023.1 A refresh of the helpful content classifier, and the last standalone one: from March 2024 the system was folded into core ranking rather than run as a separate update.
October 2023 Spam Update
Ran 4–19 October 2023, overlapping the October 2023 core update (5–19 October).1 A general spam update across multiple languages, with no specific policy category named. It is often described as having targeted scaled content abuse, site reputation abuse and expired domain abuse, but those three policies did not exist yet: Google announced them in March 2024.
November 2023 Core Update
Ran 2–28 November 2023, alongside a separate November 2023 reviews update (8 November – 7 December).1 Review signals moved into Google’s regular core ranking work rather than continuing as periodic standalone updates.
2024
March 2024 Broad Core Update and Spam Update
Two distinct updates launched the same day, 5 March 2024. The core update ran a full 45 days; the accompanying March 2024 spam update ran about 15 days.1
The core update incorporated the Helpful Content classifier directly into the core algorithm. Alongside it Google introduced three new spam policies: scaled content abuse, site reputation abuse (third-party parasite SEO), and expired domain abuse. Scaled content abuse is the one most often misdescribed: it targets producing content at scale to manipulate rankings “no matter how it’s created”, so it is not an AI-content policy and hand-written content at scale is equally in scope.
The combination produced large-scale ranking changes, including significant drops for sites that had published undifferentiated content at volume.
June 2024 Spam Update
Ran 20–27 June 2024. Targeted violations of Google’s spam policies across all languages and regions. Google confirmed the update was not link-specific, distinguishing it from dedicated link spam updates.
August 2024 Broad Core Update
Described by Google as correcting over-corrections from March. Several sites that had dropped in March saw partial or full recovery. Ran 15 August – 3 September 2024, about 19 days.1
November 2024 Core Update
Ran 11 November – 5 December 2024 (approximately 24 days), longer than the standard two-week rollout. Google described the goal as continuing to surface genuinely useful content and reduce content produced primarily to rank. Third-party tracking characterised it as less disruptive than March and August.
December 2024 Core Update
Ran 12–18 December 2024, launching within a week of the November update completing. Google confirmed the two updates addressed different core systems. More volatile than November despite the shorter duration, though less disruptive than March and August.
December 2024 Spam Update
Ran 19–26 December 2024. A general spam update across all languages and regions; not link-specific.
2025
Three broad core updates and one spam update in 2025, fewer than 2024 (four core updates, three spam updates).
March 2025 Core Update
Ran 13–27 March 2025 (14 days). Google’s stated aim was to better surface relevant, satisfying content across all site types. No specific content categories identified as targets. The two-week rollout did not translate into large effects: Semrush put it within “a mere .1 difference” of the December 2024 update against baseline volatility, Similarweb called it “not the most volatile core update” it had seen, and SISTRIX found no major rising or falling domains in the top 100.
Full analysis: what the tracking tools measured, and which UK domains moved.
June 2025 Core Update
Ran 30 June – 17 July 2025 (approximately 17 days). Produced notable volatility, with intense ranking movement between 11 and 14 July. Semrush recorded over 16% of post-update top-ten URLs as having ranked outside the top twenty beforehand, the highest churn in the four years it had tracked the metric and 72.4% above the March 2025 update. Some sites that had dropped in earlier updates saw partial recovery.
Full analysis: top-ten churn, the vertical split, and the helpful content recoveries.
August 2025 Spam Update
Ran 26 August – 22 September 2025 (approximately 27 days). A general spam update across all languages and regions, with no target named by Google. Impact was felt quickly, within 24 hours of launch, with a second intensity spike observed around 9 September. Aggregate SERP movement stayed low: SISTRIX measured most of it below normal daily noise, consistent with a penalty-style update that removes a few domains outright rather than reordering many.
Full analysis: why the tracking data and the practitioner reports disagree.
December 2025 Core Update
Ran 11–29 December 2025 (18 days). No specific content categories or site types identified as targets. Volatility was concentrated at weekends, with major ranking movement on 13 and 20 December. In SISTRIX UK data Wikipedia was the largest absolute loser, down 456.69 visibility points to 7,775.59, though that is only a 5.55% fall against its very large base. Thesaurus.com was the largest absolute winner, up 55.55 points (24.32%).
Full analysis: winners, losers, and the intent-specificity signal.
2026
February 2026 Discover Update
Ran 5–27 February 2026. The first Discover-specific update Google has officially confirmed.7 Unlike broad core updates, it applied to the Discover feed rather than Search rankings. Initially limited to English-language users in the US, with a stated intention to expand globally.
Stated goals: more locally relevant content from sites based in the user’s country; less sensational and clickbait content; more in-depth, original, and timely content from sites with demonstrated expertise.
Google’s own illustration of how the expertise signal works is topic-by-topic rather than site-wide: “a local news site with a dedicated gardening section could have established expertise in gardening, even though it covers other topics. In contrast, a movie review site that wrote a single article about gardening would likely not.”
Full analysis: the local expertise mechanic, and why a Discover-only update matters.
March 2026 Spam Update
Ran 24–25 March 2026, a rollout of 19 hours and 30 minutes and the shortest confirmed spam update on the dashboard. A general spam update with no stated specific targets. Google confirmed only what it excluded: link spam and site reputation abuse were not in scope.
Full analysis: why the named targets attributed to it are inference.
March 2026 Core Update
Ran 27 March – 8 April 2026 (12 days). Standard boilerplate from Google: a regular update to better surface relevant, satisfying content. Global and UK readings diverged sharply. SE Ranking recorded 79.5% of top-three URLs changing position against 66.8% in December 2025, while SISTRIX’s UK radar logged no major movements at all. The spam update completing one day earlier also confounds the comparison with December.
Full analysis: the volatility data, and why the two readings disagree.
May 2026 Core Update
Ran 21 May – 2 June 2026, a rollout of 11 days and 21 hours. Volatility came in distinct waves rather than at the start and end: 23 May, again around 30 May, and a further spike on 2 June hours before Google confirmed completion. Practitioners described it as heavier than March, though no provider published a figure comparing the two. The rollout coincided with Google I/O, during which Google made Gemini 3.5 Flash the default model in AI Mode, expanded AI Mode, and announced a major search interface redesign. That timing complicates attribution: ranking shifts during the rollout period may reflect the core update, the AI Mode expansion, or the underlying model change.
Full analysis: the rollout, and the attribution problem Google I/O created.
June 2026 Spam Update
Ran 24–26 June 2026 (approximately two days). The second spam update of the year, a general improvement to SpamBrain with no specific policy category named, applying globally and to all languages. It ran under an existing policy that Google had clarified covers attempts to manipulate generative AI responses, which brings AI Overviews and AI Mode inside the same enforcement as the standard results. A Cornell Tech preprint published around the same time shows the limits of that: roughly 13 words of planted text on a page AI systems already retrieve inserted an attacker’s chosen entity into the generated answer in 38% to 51% of sessions, rising to 42% to 62% when spread across several pages.
Full analysis: spam enforcement extending to AI Overviews and AI Mode.
August 2026 Spam Update
Ran 18–21 August 2026, a rollout of 2 days and 16 hours and the longest of the three 2026 spam updates. Global and all languages. Announced only as a dashboard note: no Search Central blog post, no new spam policy type, and no revision to the spam updates documentation since December 2025. Google has not said whether it strengthens enforcement of the generative-AI manipulation policy the June update ran under.
Full analysis: what Google did not say, and how the rollout compared with March and June.
Patterns across updates
Several consistent themes across Google’s named updates:
Quality over quantity. Every major content update (Panda, Helpful Content, March 2024) has penalised volume-focused content strategies and rewarded depth and genuine usefulness.
Manipulation gets devalued, not penalised. Since Penguin 4.0, spammy links are devalued rather than triggering manual penalties. The shift is from reactive penalty to proactive neutralisation.
E-E-A-T has grown in importance. From the Medic update onwards, demonstrable expertise and first-hand experience have become more explicitly relevant, particularly for YMYL content.
AI content is not intrinsically penalised. Google has repeatedly stated that AI-generated content is not penalised if it is helpful and high-quality. What the March 2024 update targeted was scaled, undifferentiated AI content produced without editorial judgement.
Page experience is table stakes. Core Web Vitals are a confirmed ranking factor but rarely the primary explanation for a ranking drop. Fix them, but look to content and links first.
Footnotes
-
Google Search ranking updates — Google Search Status Dashboard. Google’s own record of confirmed ranking updates, with launch dates and rollout durations. ↩ ↩2 ↩3 ↩4 ↩5 ↩6
-
Google Says The Top Three Ranking Factors Are Content, Links & RankBrain — Search Engine Roundtable ↩
-
Google passage ranking now live in US English search results — Search Engine Land ↩
-
Introducing INP to Core Web Vitals — Google Search Central Blog ↩
-
What creators should know about Google’s August 2022 helpful content update — Google Search Central Blog ↩
-
Google’s February 2026 Discover Core Update — Google Search Central Blog ↩