Google Algorithm Updates: The Official History and Its Gaps

Key Takeaways
– Google’s published ranking log covers 3 November 2021 onward, so earlier history is reconstruction
– 40 named updates sit on the dashboard, plus 2 entries logged as ranking incidents
– Core update rollouts ran 6 days 4 hours at the shortest and 45 days at the longest
– 12 of the 16 recorded core updates finished within 11 to 20 days
– Panda joined core ranking in 2015, Penguin in 2016 and the helpful content system in March 2024
– Google states that unannounced core updates happen between the announced ones
– No announced reviews update has appeared since November 2023
– Of 20 pre-2021 entries, Possum and Fred have no Google confirmation, and “Fred” began as a joke
– BERT is the rare case where Google published a scale figure, at one in 10 US English searches

Prepared by the SearchEngineInfo Editorial Team. Editorial review completed 17 October 2026. Update names, dates and durations were read directly from Google’s Search Status Dashboard and ranking documentation on that date. Next scheduled review: 17 January 2027. Google adds entries continuously, so check the dashboard for anything after the last date shown here.

Google’s Own Record Starts in November 2021

Timelines of Google algorithm updates routinely run from 1998 to the present. Google’s published log does not.

The Search Status Dashboard ranking history holds 42 entries, the earliest dated 3 November 2021 and the most recent 24 September 2026. Google’s former ranking-updates page on Search Central now redirects to that dashboard, which makes it the current official source. Most published timelines still cite the old location.

Two consequences follow, and both matter when you cite a date.

Everything before November 2021 rests on Google’s individual blog announcements, on what Google says today about systems it launched years ago, and on third-party observation of ranking volatility. Those sources vary in reliability. A date from the dashboard and a date from a 2011 industry blog post are not equivalent evidence, and treating them as one list hides that.

The dashboard also records duration, which almost no timeline reproduces. That data answers the question site owners ask during a rollout.

Two Entries Are Incidents, Not Updates

Two dashboard entries read “Ranking is experiencing an ongoing issue” rather than naming an update. One began 15 August 2024 and ran 4 days 11 hours. The other began 5 October 2023 and ran 26 days. Both overlapped announced updates.

Counting them as algorithm updates inflates the total. The dashboard holds 40 named updates and 2 logged incidents.

The Official Record, November 2021 Onward

Names, start dates and durations below come from Google’s dashboard as read on 17 October 2026. Google reports a start date and an elapsed duration for each entry rather than an end date. Our coverage of the May 2026 core update goes into what that particular rollout changed.

Core Updates

Spam Updates

Reviews and Product Reviews Updates

The November 2023 reviews update is the most recent announced entry in this family. Nothing has followed it on the dashboard in nearly three years.

Helpful Content, Discover and Page Experience

Three announced helpful content updates appear, the last in September 2023. Google documents the helpful content system as part of core ranking from March 2024, which accounts for the absence of later entries. The page experience update for desktop is the only entry in its family on the dashboard, and our guides to Core Web Vitals and how Google uses them as a ranking factor cover what it introduced.

How Long Rollouts Take

Google publishes the duration of each rollout, and that figure answers the question site owners ask during one.

Core updates cluster. Twelve of the sixteen finished between 11 and 20 days, with a median of approximately 14.5 days. Three of the sixteen lasted longer than three weeks, while one finished in less than a week. The March 2024 core update was the longest recorded rollout at 45 days, substantially exceeding the other fifteen.

Spam updates do not cluster. They range from 19 hours 30 minutes to 29 days with no pattern worth planning around. Predicting a spam rollout’s length from previous ones has no support in this data.

The 45-day outlier coincided with two documented events. The March 2024 core update began the same day as the March 2024 spam update, and Google documents the helpful content system becoming part of core ranking in that month. Google has not stated that either event caused the extended rollout, so treat the coincidence as context rather than explanation.

What the Record Shows About Announcement Frequency

Counting the 40 named updates by year and by family shows a consolidation that matches what Google documents about its systems. Figures below are our tally of the dashboard entries.

The families narrowed. Google announced updates across five families in 2022, counting core, spam, product reviews, helpful content and page experience. In 2025 it announced two. That tracks the absorption of the helpful content system into core ranking and the absence of announced reviews updates after November 2023.

Spam updates now outnumber core updates. Four of the seven entries in 2026 through September are spam updates, against two core updates. In 2023 the ratio ran the other way, with four core updates and one spam update.

2025 had the fewest announced updates of any complete calendar year on the record, with four against ten in 2022. The comparison excludes 2021 and 2026, which the dashboard covers only in part. Reading the figure as less change would be a mistake, since Google documents unannounced core updates and the count measures announcements rather than activity.

That last point constrains how far this table can be pushed. It describes Google’s communication, not the rate at which Google changes ranking.

Systems and Updates Are Different Things

Google documents ranking systems separately from ranking updates. Systems are the ongoing technologies. Updates are announced changes to them. Collapsing the two produces the most common error in algorithm-history writing, which is describing a site as hit by a system that stopped existing separately years ago. For what carries weight across those systems today, see our analysis of Google ranking factors that matter.

Systems Google Documents as Absorbed Into Core Ranking

Attributing a current ranking drop to a new, standalone Panda or Penguin update is misleading. Google integrated both into core ranking, while the underlying content-quality and link-spam considerations remain relevant.

Systems Google Documents as Currently Running

Google’s guide to ranking systems lists these as in use. The grouping below is ours; the systems and descriptions are Google’s.

Language understanding: BERT, which Google describes as helping it understand how word combinations express meaning and intent. RankBrain, which relates words to concepts even when a page lacks the exact query terms. Neural matching, which matches concepts in queries to concepts in pages. MUM, which Google describes as used for specific applications rather than general ranking. Passage ranking, which evaluates individual sections of a page. Our explainer on how Google’s NLP, BERT and MUM work covers the mechanics, and semantic SEO covers what they mean for content.

Link and authority: Link analysis systems and PageRank, which Google states remains part of core ranking.

Quality and spam: Spam detection systems including SpamBrain. The reviews system, which Google describes as rewarding high-quality expert reviews. Original content systems, which favor original reporting over pages citing it. Reliable information systems. Removal-based demotion systems.

Result shaping: Deduplication systems. The site diversity system, which generally limits results to two listings from the same site. The exact match domain system, which limits credit for domains built to match queries. Freshness systems. Local news systems. Crisis information systems.

Note that the reviews system appears here as current even though its last announced update was November 2023. A system running without announced updates is the normal state, which is the point of the distinction.

Before November 2021

Google publishes no log for this period, so the entries below carry an evidence column rather than sitting in the same table as the dashboard data. Dates are assembled from Google announcements where they exist and from industry records where they do not, and the column says which applies to each row.

How to Read the Evidence Column

Pre-Dashboard Updates and Systems

Where the Evidence for Each Row Comes From

The 20 historical entries are classified by the type of evidence supporting them. Eight rows contain direct links to original Google announcements, while other entries rest on Google’s current ranking-systems documentation, on official statements reported by third parties, or on historical SEO observation. Some entries have more than one supporting source, so these categories should not be added together as mutually exclusive counts. Where no direct Google announcement is available, the evidence column identifies the basis for the claim.

Documented by Google today. Panda, Penguin, Penguin 4.0, Hummingbird, RankBrain, the exact match domain system and BERT appear in Google’s current ranking systems guide, which supplies their status and, for Hummingbird and the core-ranking absorptions, the dates. BERT also carries its own announcement link in the table.

Announced by Google at the time. Six now link to the original Google post: page layout, Pirate, HTTPS as a signal, intrusive interstitials, the speed update and page experience. Two do not. We found no archived Search Central post for the June 2013 Payday Loan change, which Google communicated through other channels, and no single post covering the October 2020 BERT expansion.

Confirmed by Google under an industry name. The mobile-friendly update links to Google’s February 2015 announcement. Pigeon has no Google post, because Google gave statements to reporters instead and said it had no internal name. The August 2018 core update was announced through Google’s Search Liaison account rather than the blog, so contemporaneous Search Engine Land reporting is the record.

Industry-named with no Google confirmation. Possum and Fred. The evidence is SEO observation and reporting from the time, which is why both are labelled that way rather than given a date we present as established.

Announcement Dates and Effective Dates Differ

Retrieving the original posts surfaced something most timelines flatten. Google announced four of these changes well before they took effect, and timelines that list a single date pick one convention without saying which.

Each gap runs two to seven months. A timeline entry reading “Speed update, January 2018” and one reading “Speed update, July 2018” describe the same change, and both dates can be valid once identified as the announcement date and the rollout date respectively. Our table gives both for these four rows.

The dashboard standardizes the recorded rollout start date, which narrows the ambiguity. Announcements can still precede a rollout without appearing in the log.

What Google Said About BERT

BERT is worth quoting because Google published a figure, which is rare for a ranking change. In its announcement on 25 October 2019, written by Google Fellow Pandu Nayak, Google said BERT “will help Search better understand one in 10 searches in the U.S. in English” and described the change as representing “the biggest leap forward in the past five years.” Google extended it to nearly all English-language queries the following year.

Compare that against the dashboard era, where Google publishes names and durations but no impact figures. Scale claims from Google have become rarer as the announcements have become more regular.

The Naming Problem

Five rows above carry names Google never chose, and they sit at three different evidence levels. Lists that present them as equivalent entries lose that distinction, which matters when you put one in a client report.

“Mobilegeddon” named an update Google announced in advance. Google published the mobile-friendly update roughly two months before it launched on 21 April 2015. The industry supplied the nickname, and the underlying change is documented.

“Pigeon” named a change Google described but would not announce. Google told reporters in July 2014 that it had no specific update to announce, while describing the change as tying local results closer to its web search signals. It also said it would probably not detail future changes to local search algorithms. Google published no post and confirmed it had no internal name, so Search Engine Land supplied one the day after launch. This is the weakest of the three confirmed rows, and the article labels it that way.

“Medic” named a confirmed core update, and the name misleads. Danny Sullivan announced a broad core update through Google’s Search Liaison account on 1 August 2018. SEOs called it Medic because health and finance sites moved heavily. On 14 August, Sullivan stated it was a broad core update involving all searches, which contradicts the health-specific reading the name implies. Google’s recovery guidance was that no fix exists beyond continuing to build good content.

“Possum” has no Google confirmation. SEOs observed local ranking changes in September 2016 and named them. Google said nothing on the record.

“Fred” began as a joke. Asked what to call a series of unconfirmed changes, a Google employee suggested every update be called Fred. The name attached to a March 2017 change Google never announced, and sources disagree on whether Google later acknowledged it.

Mobilegeddon, Pigeon and Medic therefore point at real, acknowledged changes under borrowed names. Possum and Fred do not. If you reference any of them, describe what changed and cite Google’s statement where one exists.

The Official Record Is Partial by Google’s Own Account

Google’s guidance on core updates states that smaller, unannounced core updates can produce gains between the major announced ones.

So the dashboard records announced updates rather than all updates. Chasing a named cause for every movement therefore wastes effort, because no published list can account for all of it.

Three practical consequences:

Absence of an entry proves nothing. Ranking movement without a corresponding dashboard entry may reflect unannounced ranking changes, competitive movement, technical issues on your own site, seasonal demand or shifts in search behavior. The absence of a logged update does not establish the cause.

Third-party volatility trackers measure something real and name it speculatively. The movement they detect may be genuine while the label attached to it has no Google source.

Dates in the dashboard are reliable, and completeness is not claimed. Use it to confirm what Google announced and when, rather than as a full account of every change.

What Google Says to Do About a Core Update

Google’s core updates guidance is the primary source for this, and its advice differs from most recovery content.

Core updates do not target sites. Google describes them as “significant, broad changes to our search algorithms and systems” that “don’t target specific sites or individual web pages,” and states that “most sites don’t need to worry about core updates.”

Read that precisely. A core update is a broad change rather than an action directed at your website, and its effects can still concentrate in particular sites, page groups or query categories. A site-level decline does not, by itself, establish a penalty or a manual action. Check Search Console’s manual actions report if you suspect one, since that is where Google reports them.

A drop does not establish that something is broken. Google describes constant change producing both gains and drops, and compares reordering to a list of restaurants where places that move down are not necessarily bad.

Wait before measuring. Google advises confirming the rollout has finished, then waiting at least a full week before comparing Search Console data. Against a median core rollout of approximately 14.5 days, that puts a first honest read at roughly three weeks from the start date. Our Search Console guide covers the date-comparison tool for that check.

Do less than you want to. For a small position drop Google advises against drastic action and against changing content that already performs well. Deleting content is described as a last resort. For a sustained drop, Google points to a whole-site assessment against its guidance on helpful, people-first content, which is where our content audit process and E-E-A-T guide apply.

Recovery runs on a longer clock than rollout. Google says some changes can take effect in a few days, while it “could take several months for our systems to learn and confirm” that a site produces helpful content over time. With no effect after a few months, Google says that “could mean waiting until the next core update.”

Read against the duration table, that is the useful comparison. Among the sixteen recorded core updates, the median rollout lasted approximately 14.5 days. Recovery follows a separate timeline: improvements may be reflected sooner, while some changes can take months or become more apparent after a subsequent core update.

How to Investigate a Traffic Drop During an Algorithm Update

The data above becomes useful when a drop arrives. Six steps, in order, and the first two prevent most wasted work.

1. Confirm an update ran, and when it finished. Check the dashboard for an entry covering your drop. Note the start date and the duration, since a rollout still in progress produces movement that has not settled. An entry that ended before your drop began makes a direct temporal association less clear, without ruling out delayed or continuing effects.

2. Wait before measuring. Google advises confirming the rollout has finished, then waiting at least a full week before comparing Search Console data. Against a median core rollout of approximately 14.5 days, that puts a first reliable comparison around three weeks from the start date. Measurements taken during a rollout are provisional and should be reassessed once it completes.

3. Segment before concluding. Compare periods in Search Console by page, query, device and search appearance rather than reading the site total. A sitewide figure hides the shape of the change, and the shape is the diagnosis. One template falling while the rest hold steady points somewhere different from a uniform decline across every page. Our Search Console guide covers the comparison tools for this.

4. Rule out your own site first. Check indexing and crawl reports, recent deployments, template changes, redirect edits and robots.txt history for the same window. A coinciding update makes a convenient explanation for a drop your own release caused. Our technical audit checklist covers what to check.

5. Look at what replaced you. For the queries that lost position, examine the results now ranking above you. Core updates reorder results, and the pages that gained may reveal differences in relevance, usefulness, format or other competitive factors. They do not establish why Google reordered the results. That comparison still beats working from a generic list of ranking factors.

6. Document before you change anything. Record the affected queries, pages and dates before editing. Google advises against drastic action on small drops and describes deleting content as a last resort. Without a written baseline you cannot tell later whether a change helped, recovered on its own, or coincided with the next update. Our content audit process covers the assessment Google points to for sustained drops.

Steps 1 and 4 together help distinguish update timing from site-specific technical causes, which is the distinction most suspected algorithm hits turn on.

Frequently Asked Questions

How far back does Google’s official algorithm update history go?

Google’s Search Status Dashboard ranking history begins on 3 November 2021. Earlier updates are documented in individual Google announcements and in third-party reporting rather than in a published Google log.

How many Google algorithm updates are on the official record?

The dashboard held 42 entries when read on 17 October 2026. Forty are named updates. Two are logged as ranking incidents titled “Ranking is experiencing an ongoing issue” rather than named updates.

How long does a Google core update take to roll out?

Of the 16 core updates on Google’s record, 12 finished between 11 and 20 days, with a median of approximately 14.5 days. The shortest was December 2024 at 6 days 4 hours. The longest was March 2024 at 45 days.

Can my site still be hit by Panda or Penguin?

Google documents Panda as part of core ranking since 2015 and Penguin as integrated into core ranking in 2016. Neither operates as a separate system that produces a discrete hit. A ranking drop may be real while that label has no current mechanism behind it.

What happened to the helpful content system?

Google documents it as part of core ranking from March 2024. Three announced helpful content updates appear on the dashboard, the last in September 2023, and none since.

My rankings moved but there is no update on the dashboard. What happened?

Google states that smaller core updates roll out without announcement. Movement with no dashboard entry is consistent with that, so the absence of a named update does not mean nothing changed.

When should I measure the impact of a core update?

Google advises confirming the rollout has finished, then waiting at least a full week before comparing Search Console data. Given a median rollout of approximately 14.5 days, a first reliable comparison lands around three weeks after the start date.

How long does recovery from a core update take?

Google says some changes can take effect within days, while confirming long-term content quality “could take several months.” If nothing changes after a few months, Google says that could mean waiting for the next core update.

Why have there been no reviews updates recently?

The last announced entry in that family is the November 2023 reviews update. Google still lists the reviews system among its current ranking systems, so the system continues to operate without announced updates.

Which Google algorithm updates did Google never confirm?

Among the widely cited pre-2021 names, Possum from September 2016 and Fred from March 2017 have no Google confirmation. SEOs observed ranking movement and named it. The name Fred began as a joke from a Google employee who suggested calling every unnamed update Fred.

Was Mobilegeddon a real Google update?

The update was real and Google announced it roughly two months in advance, launching the mobile-friendly update on 21 April 2015. Google never used the name Mobilegeddon, which the SEO industry supplied. The same pattern applies to Medic in August 2018, which named a broad core update Google did confirm.

Limits of This Guide

The dashboard is announced updates, not all updates. Google documents unannounced core updates, so no published list can be complete.

Durations describe rollouts, not effects. A duration tells you when Google finished shipping a change. It does not tell you when your rankings stabilized.

Pre-2021 dates carry mixed evidence, which is why each row is labelled. Google’s statements about Panda, Penguin, Hummingbird, RankBrain, BERT and the exact match domain system are what Google says today about systems it launched earlier. Dates for announcements Google no longer indexes prominently come from industry records of those announcements. Possum and Fred rest on SEO observation with no Google confirmation, and we have marked them rather than dropping or promoting them.

Day-level precision is not claimed for most pre-2021 entries. Where a row gives a month, that is the resolution the available records support. Three rows include day-level dates: the mobile-friendly update on 21 April 2015, the broad core update on 1 August 2018, and the BERT announcement on 25 October 2019. The first and third link to Google announcements, and the August 2018 date comes from Google’s public communication through its Search Liaison account. Hummingbird’s August 2013 comes from Google’s current documentation at month resolution.

Our groupings are editorial. Google lists its ranking systems without the categories we used to organize them. The system names and descriptions are Google’s; the arrangement is ours.

The record grows. Figures here were read on 17 October 2026. Any update after 24 September 2026 is absent, and the counts and medians shift as Google adds entries.

This covers Google only. Other search engines publish no comparable log. For how they source and rank results, see our guide to alternative search engines and which maintain their own indexes.

Methodology

How the figures in this guide were produced, so you can reproduce or challenge them.

Counting. We read Google’s Search Status Dashboard on 17 October 2026 and counted named ranking updates separately from entries logged as ranking incidents. That separation gives 40 named updates and two incidents across 42 entries. Counts that treat all 42 as updates will differ from ours by two.

Durations. Rollout statistics come from the start dates and elapsed durations Google publishes per entry. The median is the mean of the eighth and ninth values in the sorted list of 16 core-update durations. Ranges are the minimum and maximum of that list.

Annual and family tallies. We assigned each named update to a calendar year by its start date and to a family by its official name. Row totals and the grand total were checked against the 40-update count. Comparisons between years exclude 2021 and 2026, which the dashboard covers only in part.

Pre-November 2021 entries. Google publishes no log for that period. We reconstructed those rows from Google’s current documentation, from archived Google announcements we located and verified by page title and status, and from historical industry reporting where no Google record exists. Each row carries an evidence label, and rows with no Google link say so.

Announcement dates against effective dates. Where Google announced a change before it took effect, we give both dates rather than choosing one convention silently.

What we did not do. We did not read the body text of every archived announcement, since several sit behind more navigation than our tools retrieve; those are cited as announcements rather than quoted. We did not use third-party volatility tracker data as evidence for any entry. We did not extend the table before 2011.

Sources and References

Primary Google documentation

• Search Status Dashboard, ranking history (Google)
• A guide to Google Search ranking systems (Google Search Central)
• Google Search core updates and your website (Google Search Central)
• Understanding searches better than ever before, the BERT announcement (Google, 25 October 2019)

Archived Google announcements for pre-dashboard changes

• Page layout algorithm improvement (Google Search Central Blog, January 2012)
• An update to our search algorithms, the copyright removal signal (Google Inside Search, August 2012)
• HTTPS as a ranking signal (Google Search Central Blog, August 2014)
• Finding more mobile-friendly search results (Google Search Central Blog, February 2015)
• Helping users easily access content on mobile (Google Search Central Blog, August 2016)
• Using page speed in mobile search ranking (Google Search Central Blog, January 2018)
• Timing for bringing page experience to Google Search (Google Search Central Blog, November 2020)

Contemporaneous reporting, where Google published no post

• Pigeon: Search Engine Land’s name for the new Google local search update (Search Engine Land, July 2014)
• Google confirms the August 2018 core update finished rolling out (Search Engine Roundtable)
• Google states the August 2018 core update is global and spans all categories (Search Engine Roundtable)

Secondary sources, used only for pre-dashboard dates

Every name, date and duration from November 2021 onward comes from Google’s dashboard, and the year-by-year counts and duration ranges are our arithmetic on those entries, which you can recompute. The pre-2021 table is different. Google publishes no log for that period, so the dates there come from industry records of Google’s announcements and, for the unconfirmed entries, from contemporaneous SEO reporting. Those rows carry an evidence label for that reason. Where Google’s current documentation covers an older system, we cite Google rather than the industry record, and the label says so.