Google I/O 2026: What 20+ Search Leaders Are Saying

Google I/O 2026 put AI Mode at the center of Google Search, and the search industry spent the following weeks arguing about what that means for the work. Most of the people the industry listens to agree that the announcements are significant. They disagree on what to call the work that comes next, how to measure it, and how much of the old playbook still holds.

The debate did not start at the keynote. Google published its own guide to optimizing for generative AI features on May 15, four days before I/O, and several of the sharpest responses were aimed at that guide. Other positions quoted here were set out months earlier and became more relevant once I/O confirmed the direction. Each view below carries its source and date, so you can see which comments react to I/O and which ones came before it.

More than twenty search leaders have weighed in on the shift, from the Google executive who announced it to the practitioners who will implement it. For the operational side, see our 90-day playbook on the I/O reset.

What Google announced at I/O 2026

The Search announcements were published in Google's I/O 2026 Search post on May 19, 2026, written by Elizabeth Reid, VP of Search.

How leaders reacted to the keynote

Elizabeth Reid set the framing for the week. In her keynote remarks, reported by PPC Land, she said that "Google search is AI search through and through." That line did more to shape the industry conversation than any single feature demo.

Marie Haynes published one of the most detailed practitioner breakdowns the day after the keynote, in a video analysis on May 20. Her view was that the changes split the field. Some sites will lose, while those "on the cutting edge in creating agentic experiences" stand to gain. Her structural point was that SEO has always optimized for a single moment when a user types a query, and background agents push teams to optimise for continuous tracking instead.

Barry Schwartz of Search Engine Roundtable gave the community its factual baseline, reporting that Gemini 3.5 Flash now powers AI Mode as the default model worldwide, alongside the new agentic features.

Rand Fishkin of SparkToro published new zero-click data three weeks after I/O. His June 9 analysis found that 68 percent of Google searches ended without a click between January and April 2026. He advises brands to invest in what he calls zero-click marketing, building influence and awareness without needing a visit, and notes that traffic can fall sharply while revenue rises.

The debate over whether AI search is still SEO

Mike King, founder of iPullRank, drew the hardest line, and he did it four days before I/O. Responding to Google's May 15 guide, he published "Google's Guidance on AI Search is Naive and Self-Serving", arguing that it suits Google for SEOs to remain "the janitors of the web." His case is organizational as much as technical. If AI search is folded back into SEO, teams will not get the budget or skills the work needs, from retrieval to passage engineering to agent protocols.

The reaction to King's post brought in other voices. PPC Land's coverage quoted Matthew Proctor, CEO of Narrative Bent, making the pragmatic point that whatever the discipline is called, the work will keep falling on the people already responsible for search. The same piece cited Nikita Vlasyuk, CTO of DeepSEO, pointing to AI Overviews passing a billion users and more than half of Google searches ending without a click. It also referenced Jordi Ribas, Microsoft's Corporate VP of Search and AI, who has written about generative engine optimization as a new optimization discipline. Several commentators used that contrast between Google and Microsoft to argue that a Google-only strategy is too narrow, a case we also made when Microsoft published the AI search guide Google could not write.

Lily Ray, VP of SEO Strategy and Research at Amsive, holds the steadier position. Her 2025 article on the rise of AI search argued that it does not mean SEO is dead. Summarizing Google's guide on May 15, 2026, she told followers to ignore most GEO and AEO hacks, naming chunking and llms.txt.

Mordy Oberstein described the period as old SEO giving way to new SEO in a June 2025 interview with Search Engine Land, saying he feels the industry is in an era of refinement. Garrett Sussman of iPullRank opened 2026 with a post on the range of AI search opinions, noting that he agrees with many of them and disagrees with many others, a fair description of the field heading into I/O.

Measurement beyond traffic

Aleyda Solis, founder of Orainti, has been the clearest voice on measurement. In a March 2026 piece for Peec AI, she said teams should stop using traffic as a reliable KPI and use a blend of branding and performance measures, naming AI visibility, sentiment, purchases, and revenue. Her February analysis of SERP shifts showed paid results taking a sharply larger share of the clicks that remain in some categories.

Kevin Indig, author of Growth Memo, has published user research on how people behave inside AI features. His May 25 study on AI Overviews compared with AI Mode found that users in AI Mode tend to accept the answer, and his earlier research found that most users clicked nothing at all. For brands, the measurable goal shifts toward being retrieved and cited inside the answer.

Wil Reynolds, founder of Seer Interactive, has argued that visibility alone was never the goal. At SEO Week 2025 he told the audience that Seer's organic traffic was near its lowest point while revenue was near its highest. Seer has since built that thinking into a measurement framework called Be Seen, Be Believed, Be Chosen, written up by John Lovett, Seer's VP of Analytics.

The case for keeping SEO fundamentals

Cyrus Shepard of Zyppy offers a useful check on the idea that everything has changed. His May 2026 analysis of AI citation ranking factors scored how well a URL ranks for the exact query second only to whether the URL is accessible to crawlers. His summary is that winning at SEO wins AI citations most of the time, with extra steps.

Glenn Gabe of G-Squared Interactive keeps documenting what he calls Mt. AI, the pattern of sites that surge on scaled AI content and then collapse. His April 2026 case study followed a site that grew during the March 2026 core update and was later removed from the index after a manual action for scaled content abuse.

Britney Muller made the point in one line in a 2025 LinkedIn post, telling the industry it is rediscovering SEO by way of "AI SEO" or GEO. Nathan Gotch, founder of Gotch SEO, wrote after the 2025 AI Mode launch that brands should invest in brand and storytelling, diversify where they appear, and engineer for AI Mode without overfitting to it.

Retrieval, structure and entities

Andrea Volpini, CEO of WordLift, argues that AI search depends on memory, structure, and connection. SEO Week's recap of his April 2026 talk summarized his view that structured data affects retrieval accuracy and that search is becoming navigation as well as answers. WordLift's March 2026 article on AI memory and entity hubs reported answer accuracy improving by up to 29.8 percent in its tests.

Dan Petrovic of DEJAN works on the shift through information retrieval. His March 2026 research on grounding snippets found that Google's AI uses a fixed budget of roughly 2,000 words of source text per query, with the top source receiving the largest share. The practical lesson is to write and test content at the passage level as well as the page level.

Gianluca Fiorelli of ILoveSEO raised a risk that the keynote optimism skipped. In his 2025 piece on AI Mode becoming Google's default interface, he warned that memory and hyper-personalization could create a search echo chamber. Personal Intelligence, announced at I/O 2026, makes that warning more pressing, because familiar brands and strong entities stand to benefit when answers adapt to what a user already knows.

What enterprise teams should change after I/O 2026

Behind the argument over names, these leaders describe the same mechanics. Rankings and technical health still set the floor, structured data and clear entities now carry more weight, and traffic no longer shows on its own whether search is working. The split is over Google's framing. One camp reads I/O as continuity with SEO, while King and others read Google's guidance as a platform protecting its own surface as the work spreads across engines that do not share Google's index. We sit closer to the second camp, and we explained why in Google's guidance is not the whole map.

For an enterprise team, the useful response is a set of practical changes. Rebuild measurement around AI citations, branded demand, and revenue, so a fall in sessions does not hide growth or decline. Keep the ranking and technical fundamentals intact, since Shepard's data shows they still drive citations. Write and structure content so individual passages answer specific questions, and treat visibility in ChatGPT, Perplexity, Copilot, and Gemini as part of the search scope. Our explainer on generative engine optimization covers how that work fits with SEO.

If you want this turned into a prioritized roadmap for your own domains, with measurement rebuilt around AI citations and the retrieval work scoped for your category, request an audit or get in touch. We run AI search audits for retainer and project clients.