Google algorithm updates: the full history, 2003 to today

By James Burfield · Last reviewed 1 October 2026

Every confirmed Google update in one place: what it was, when it ran, and what the data showed. Dates come straight from Google’s Search Status Dashboard, not from memory or social media threads. I update this page after every confirmed rollout.

Where things stand right now

As of 1 October 2026, the September 2026 spam update is still rolling out. It started on 24 September, putting it more than six days in, which is within Google’s stated window of up to two weeks. This follows the August 2026 spam update, which ran from 18 to 21 August and lasted 2 days and 16 hours.

The last core update was May 2026, completed on 2 June. That leaves the core systems about four months without a refresh. Two or three core updates a year is a recent habit, and Google can change that cadence whenever it wants.

2026 updates

Notes on 2026 so far

The September spam update is the fourth this year, following March, June and August. Each rollout has been longer than the last: under a day, two days, two days and 16 hours, and now more than six days. The gaps between them have shrunk from 91 days to 55, then 36. Google has published no explanation. The news post has the detail.

The August spam update arrived on 18 August after a 7 week gap with nothing confirmed, and finished on 21 August. Spam updates have been the quick ones this year: March took under a day, June just over 2, this one 2 days and 16 hours.

The February Discover update was the first update Google has confirmed specifically for Discover. If a chunk of your traffic comes from Discover, this was the one to watch: reports pointed at clickbait headlines and recycled content losing visibility. Publishers felt it far more than classic search-first sites.

The March spam update rolled out in under 20 hours, unusually fast for a spam update. The March core update followed 3 days later and ran a fairly typical 12 days.

A caution on the March core: some blogs claimed it introduced composite Core Web Vitals scoring, where failing one metric compounds penalties. Google announced nothing of the sort, and I’ve seen no data supporting it. Treat that claim with suspicion; it’s a good example of why this site keeps a who to listen to page.

The May core update was described by Google as a regular core update. Rollout was clean at just under 12 days, and volatility looked ordinary compared with the 2023 and 2024 monsters.

2025 updates

Notes on 2025

A quiet year by recent standards: 3 core updates and 1 spam update. The June core is the one the community associates with MUVERA, Google’s multi-vector retrieval research. The timing fits and the retrieval behaviour shifted in ways consistent with it, but Google never confirmed the link, so file that under informed speculation rather than fact.

The December core landed 2 weeks before Christmas. Google spent years avoiding holiday-season updates, then dropped that courtesy around 2024. Assume nothing is sacred in the calendar any more.

2024 updates

Notes on 2024

The March 2024 core update was the biggest core update in years, and the messiest. It folded the helpful content system into the core ranking systems, ran for 45 days, overlapped a spam update, and came with a wave of manual actions that deindexed entire sites overnight for scaled content abuse. Google claimed the combined changes cut low-quality, unoriginal content in results by around 45%.

The August core came with unusual messaging: Google said it aimed to better surface small and independent sites, a direct response to the outcry from publishers flattened by the September 2023 helpful content update. Some of that cohort saw partial recoveries here and again in November. Most did not.

November and December brought 2 core updates and a spam update inside 6 weeks, the tightest cluster of confirmed updates Google has ever shipped. If your rankings moved in Q4 2024, untangling which update did it is genuinely hard.

2023 updates

Notes on 2023

The September helpful content update is the most consequential single update on this page. A sitewide classifier hammered thousands of small independent publishers, many of whom had readable, well-researched content. Forums, Reddit and big brands gained much of what they lost. It was also the last standalone HCU: the system merged into core updates in March 2024, which means there was never a clean ‘HCU refresh’ for hit sites to recover in.

The November reviews update was the last announced one. Google said the reviews system now runs at a regular, ongoing pace and stopped confirming its updates. So any reviews-style volatility since then is invisible on official channels.

October and November each had overlapping updates running at the same time. 2023 is where ‘which update hit me?’ stopped having tidy answers.

2018 to 2022: the broad core era

This is when Google moved to named, confirmed broad core updates on a semi-regular cycle, and when E-A-T (later E-E-A-T) entered every SEO conversation.

2003 to 2017: the classic era

Older updates, condensed. Most named updates from this period were later absorbed into core systems: Panda and Penguin both run inside core ranking today.

How to tell if an update actually hit you

• Check the dates first. Compare clicks and average position in Search Console for the rollout window against the weeks before, matching days of the week. If your drop started 3 days before the rollout began, an update didn’t cause it.
• Segment before you panic. Split by page type, topic and intent. A uniform sitewide drop suggests a classifier or a technical problem. One template or topic cluster falling points you at the actual issue.
• Check rankings, not just clicks. Stable rankings with falling clicks usually means the SERP changed around you: AI Overviews, more features, different layout. That’s a different problem with different fixes.
• Volatility trackers tell you the weather. Semrush Sensor and the like confirm something is happening. They can’t tell you whether it happened to you. Your Search Console data can.

What recovery actually looks like

Some uncomfortable observations from watching this for years:

• Core update recoveries mostly arrive at later core updates. Google has said as much. Improving your site mid-cycle rarely shows results until the next core rollout reassesses you, and sometimes the one after that.
• Spam is the exception. Algorithmic spam suppression can lift once the problem is fixed and recrawled. A manual action needs a fix plus a reconsideration request, and honesty in that request matters.
• The HCU cohort is the cautionary tale. Sites hit in September 2023 waited over a year for the partial recoveries some saw in August and November 2024. Plenty of decent sites never came back. Anyone selling you a guaranteed recovery timeline is guessing.
• Sometimes there’s genuinely nothing to fix. Google’s own guidance says a drop after a core update doesn’t mean your pages are bad; relevance got re-scored. That’s cold comfort, and it’s also often true.

Method and sources

Rollout dates and durations come from the Google Search Status Dashboard (2023 onward) and Google’s announcements for earlier years. Durations are rounded to whole days. Commentary is mine and marked as such where it goes beyond what Google confirmed. Google’s own documentation on core updates and spam updates is worth reading in full; for how the systems fit together, see my guide to Google’s ranking systems.