Definition
Algorithm updates are changes to the systems a search engine or AI platform uses to crawl, rank, retrieve and present content. In classic search they mean changes to ranking, such as Google's core and spam updates. In AI search they also include model releases and retrieval changes that alter which sources an assistant cites and which brands it names. Either kind can move traffic and visibility overnight without any change on your site.
Google makes thousands of smaller changes a year and announces the larger ones on its Search Status Dashboard. In 2025 it confirmed core updates starting March 13, June 30 and December 11, plus a spam update from August 26. In 2026 so far it has confirmed a spam update on March 24, the March core update (March 27 to April 8), the May core update (May 21 to June 2), and spam updates starting June 24, August 18 and September 24, the last completing on October 8. A Discover-specific update ran in February 2026.
Core updates are broad changes to how Google assesses content and usually take about two weeks to roll out. Spam updates target practices that violate Google's spam policies. The separate helpful content system was folded into core ranking in March 2024, so 'helpful content updates' no longer appear as their own entries. Because Google's AI Overviews and AI Mode ground their answers in Google's index and ranking systems, a core update can change AI citations as well as blue-link rankings.
AI platforms add a second category. Google swaps the models behind its AI features (Gemini 3 became the default for AI Overviews in January 2026, and Gemini 3.5 Flash for AI Mode in May 2026), and each swap can change answer length, sources and the brands named. OpenAI's model releases have similar effects on ChatGPT. Our data showed average citations per ChatGPT response dropping across models around the GPT-5.3 rollout on March 4, 2026, and ChatGPT stopped exposing its search sub-queries in responses on March 5 to 6. In October 2026, after the GPT-6 release, ChatGPT links were reported to stop passing a referrer, which changed analytics without changing actual traffic. These changes usually come without the kind of announcement Google posts for core updates, which makes AI answer volatility harder to diagnose.
For SEO and GEO teams, handling updates comes down to detection and diagnosis:
- Keep a timeline of known events: Google dashboard entries, model releases and platform changes.
- Track rankings, Search Console data, AI citations and AI referral traffic separately, so you can see which surface moved.
- Compare against competitors on the same prompts and queries. If everyone moved, it was the platform; if only you moved, look at your own pages.
- Wait for a rollout to finish before acting on core update data, then compare pages that gained and lost.
- Invest in what survives updates: helpful, specific content that demonstrates E-E-A-T, clean technical foundations, and consistent coverage across the sources AI systems use.
Tracking the AI SEO impact on rankings helps separate classic ranking shifts from changes in AI answers.
Examples of Algorithm Updates
- A home improvement publisher loses 30% of Google clicks during the March 2026 core update, waits until the rollout completes on April 8, then compares its losing pages with competitors' winning pages before rewriting anything.
- A SaaS company sees its ChatGPT citations fall in the first week of March 2026 and, after checking that competitors fell too, attributes the drop to the platform-wide decline in citations per response around the GPT-5.3 rollout.
- A recipe site notices AI Overview citations shift toward video sources the week Gemini 3 becomes the default model for AI Overviews, while its blue-link rankings stay flat.
- A link vendor's client network loses visibility during the August 2026 spam update, and the affected sites recover only after removing paid links.
