The Google May core update: balancing core algorithm volatility with the rise of agentic search

Tracking the volatility spikes

Unlike some historical updates that rolled out gradually over weeks, the May update caused immediate, sharp ranking movements. According to tracking data reported by Search Engine Land, the initial impact was felt significantly on Saturday 23 May, only 48 hours after the official announcement.

A second wave of volatility occurred on Saturday 30 May, with a final burst continuing right up until the completion status was published on the Google Search Status Dashboard, as documented by Search Engine Journal. Practitioners widely described this rollout as larger and more impactful than the March update.

Timeline of key events:

• 21 May: Update announced via LinkedIn and the Google Search Status Dashboard
• 23 May: First major volatility spike (approximately 48 hours post-launch)
• 30 May: Second major volatility spike
• 2 June: Rollout complete, with a final burst of movement on the day of closure

Google's public guidance remained unchanged: there are no specific technical fixes for a core update drop. The algorithm's goal remains to surface relevant, satisfying content written for human users. If your site experienced a decline during this 12-day window, the drop indicates that Google's systems now view a competitor's content as more helpful or contextually relevant to the user's qu

The AI shift: traditional SEO meets AEO and GEO

The May core update didn't happen in isolation. It rolled out alongside significant iterations to Google's AI Overviews and the deployment of its "AI Mode" search interface, announced at Google I/O 2026 the week prior.

Google's updated AI search documentation has clarified its position on Answer Engine Optimisation (AEO) and Generative Engine Optimisation (GEO): both disciplines are still, at their core, an extension of traditional SEO. Optimising for generative AI visibility relies on the same core tenets as organic search: structural clarity, semantic depth, topical authority, and clear entity alignment. The executional demands are simply higher.

To ensure your site's content is picked up and cited within AI Overviews, you'll need a more rigorous execution of your existing content marketing SEO framework:

• Format for extraction. Structure data cleanly using bullet points, tables, and explicit definitions that AI systems can readily parse.
• Target absolute clarity. Refine your on-page SEO so content blocks lead with direct, factual answers before expanding into nuance.
• Solidify entity data. Ensure your brand's authorship and entity relationships are explicitly mapped via Schema markup, so Google's Knowledge Graph can connect the dots.
• Build external trust signals. Ensure your brand is being mentioned in the conversations your customers are already having, across forums, reviews, and authoritative publications.

The future of crawl management: why llms.txt is speculative

As site owners look for ways to interact with AI platforms, a debate has emerged around llms.txt: a proposed text file placed in a site's root directory to provide a lightweight, markdown-based map of content specifically for LLM scrapers.

The debate intensified when site owners noticed an apparent contradiction. Google Search Central documentation states no special AI files are needed for generative AI search visibility, yet Chrome's Lighthouse developer documentation recently added a check for the existence of an llms.txt file.

Google's John Mueller clarified the situation on Reddit (reported by Search Engine Journal). He stated that llms.txt is purely speculative for now. The file format has existed for years, yet no major production AI systems actively use it to alter crawl behaviour. Mueller compared it to the old keywords meta tag: AI crawlers can read your HTML directly, so a separate summary file adds little value, and server logs confirm that major AI bots don't even request the file. He noted the Lighthouse check is a developer-tooling addition and is not an endorsement from Google Search.

Instead, Google is pointing technical SEOs towards a different proposal: WebMCP (Web Model Context Protocol).

Protocol / file: llms.txt

Current status: Speculative (not adopted by major engines)

Primary function: Provides a static, text-only summary map of site content for LLM reading.

Protocol / file: WebMCP

Current status: Active origin trial (Chrome 149, May 2026)

Primary function: A W3C-backed standard that lets AI agents interact with live UI elements (e.g., pulling dynamic pricing or completing a checkout).

A note on WebMCP's current maturity: WebMCP was co-developed by Google and Microsoft and is being standardised through the W3C Web Machine Learning Community Group. It entered a public origin trial in Chrome 149 at Google I/O 2026. As of June 2026, it's available in Chrome as an early preview and shouldn't yet be treated as a fully stable, cross-browser standard. Cross-browser support (Edge, Firefox, Safari) is still on the roadmap. That said, it has clear institutional backing and a defined purpose, which is precisely why Mueller and the Chrome team are pointing to it over llms.txt

The priority: don't block AI agents

Mueller's guidance highlights the most practical challenge for modern webmasters: the biggest hurdle for most sites isn't implementing speculative files like llms.txt. It's avoiding the accidental or overly aggressive blocking of AI agents.

Many IT setups and firewalls block unknown bots by default to conserve bandwidth or protect scrapable data. As consumers shift towards using AI agents to perform comparative research, filter product listings, and complete tasks, blocking these agents removes your brand from their consideration sets entirely. This applies to both standard corporate sites and high-volume e-commerce SEO platforms.

If an autonomous agent can't access your site to read your interface, it can't recommend your product to the end user. For long-term organic visibility, maintaining open, unhindered access for verified search and AI agents is no longer optional.

Building authority in the new landscape

Building authority in this new landscape means validating that your external reputation matches your on-site optimisations. Securing links and brand mentions via proactive off-page SEO remains the baseline metric that AI discovery tools use to cross-reference your legitimacy. To track how these algorithmic shifts impact your actual visibility across traditional and AI surfaces, deploying clear, consolidated SEO reporting services is vital to separate actual performance trends from short-term volatility.