On May 6, 2026, Google’s VP of Product Management for Search, Hema Budaraju, announced five changes to AI Overviews and AI Mode. The framing was modest — “help users find original content and trusted sources.” The mechanics are not modest. AI Overviews now appear on roughly 25.8% of US searches and trigger on 39% of informational queries. Organic CTR has dropped as much as 61% in some categories. Every adjustment Google makes to how links surface inside AI responses moves real revenue between businesses.
This is the fifth significant iteration since AI Overviews launched. It is the first that materially changes how citation slots are allocated, labelled and contextualised. Brands that restructure for it now will compound visibility through the rest of 2026. Brands that wait will read the case studies of the ones that did not.
What Google Actually Changed on May 6, 2026
Five product changes ship across AI Overviews and AI Mode. Not all five carry the same weight for businesses competing for citation.
Inline links replace bottom-stacked source lists
Until now, AI Overviews placed source links in a sidebar or under the response. As of this update, Google places links directly next to the generated text they support. A bulleted point about a California bike trip can sit next to a clickable link to a guide about that route. The user reads the claim and the source attribution at the same moment.
This is the most consequential of the five changes. It rewards content structured for atomic retrieval — short paragraphs, single-claim sentences, clear citations. It penalises long, meandering prose where no individual line is the obvious source.
Hover previews on desktop
Hovering over an inline link on desktop now shows a preview card with the website name, page title, and what Google describes as additional context. The user sees where they are going before they click. Brands with strong title tags and clear branding gain. Brands with vague titles and inconsistent identity lose.
Subscription labels for users’ news subscriptions
Google now labels links from publications the user has subscribed to. In Google’s own early testing, users were significantly more likely to click links flagged with the “Subscribed” tag. Publishers can opt in via Google’s developer documentation.
For B2B brands and SaaS, this is largely irrelevant. For media properties and any business that owns a paid newsletter or subscription product on a participating platform, this is a direct CTR boost worth chasing.
“Further Exploration” suggestions at the end of responses
Many AI responses now end with a section linking out to articles or in-depth analyses on adjacent facets of the topic. Google describes them as “where to go next” suggestions. They reward depth: pillar content, technical white papers, and detailed comparison pages are eligible. Surface-level introductions are not.
Expert and community perspectives from forums
AI responses will now include previews of perspectives from public online discussions, social media, and other firsthand sources. Each perspective includes the creator’s name, handle, or community. Google has been heavy on Reddit content in standard search since 2024. Now those perspectives appear inside the AI summary itself, not just in the blue links beneath it.
This formalises forum content as a citation category. A Reddit thread, a Quora answer, or a niche industry forum post can now appear as evidence inside an AI Overview. Brands without owned community presence have a new gap to fill.
Why This Update Matters More Than the Last Three
The four AI Overviews iterations between October 2024 and February 2026 were primarily about reducing hallucination, expanding coverage, and tightening the response template. They did not move citation slots. This one does.
Three structural shifts deserve specific attention.
Citation surface area expanded. Inline links plus “Further Exploration” plus expert perspectives mean a single AI response can now legitimately cite eight to twelve distinct sources, up from the typical three to five. More citation slots means more opportunity per query. It also means more competition per slot, because the threshold to appear is lower.
Trust signals are now visible in the SERP. The “Subscribed” label introduces a visible trust marker inside an AI response. Until now, all citations looked the same. They no longer do. Once Google introduces one trust label, more follow. Verified expert profiles, awarded brand badges and source quality tiers are predictable next steps.
Forum content is a first-class citizen. For two decades, SEO best practice treated forum content as low-quality. Google’s update inverts that for queries where firsthand experience matters. The implication for brands is direct: if your CMO has not posted on Reddit, on the relevant subreddit for your industry, in 2026, your competitor’s CMO is about to.
What Changes for Citation Strategy
The update favours specific structural choices. None of them are new ideas. All of them now matter more.
Atomic paragraphs. Inline placement rewards content where one paragraph carries one claim. Three-sentence paragraphs with a clear subject, verb and supporting fact will be retrieved and cited more often than 100-word essays where the model has to guess which sentence to attribute.
Specific page titles. Hover previews surface the page title verbatim. A page titled “Services” or “About” will perform worse than a page titled “GEO Audit Methodology — 12-Week Sequence by Hyperdot.” Specificity is now a CTR lever, not just a ranking lever.
Topic clusters with depth. “Further Exploration” rewards sites that have multiple pieces on the same topic at varying levels of depth. A pillar page plus six cluster posts plus one downloadable PDF is eligible to appear three times in the same response — once in inline citation, once in further exploration, once via subscription label if applicable.
Owned community presence. Expert perspectives mean Reddit profiles, Quora profiles, Stack Overflow accounts and industry forums become extensions of the brand’s citation surface. Treating them as social channels misses the point. They are now a structured data source for AI.
What Brands Should Do This Week
Five concrete actions for any business that depends on AI search visibility for revenue.
1. Audit citation surface for your top 20 commercial queries. Run each query through ChatGPT, Gemini, Google AI Mode, and Perplexity. Record which brands are cited, where the citation appears (inline, further exploration, expert perspective), and what source page is linked. Build a baseline. The next 90 days of strategy work depend on knowing where you stand today.
2. Restructure top five service pages for atomic paragraphs. Take the most commercially valuable five URLs on your site. Rewrite them so that every paragraph contains one idea, supported by one specific fact, in one to four sentences. Bury the marketing prose. Surface the claims.
3. Activate or claim founder profiles on Reddit, Quora, and at least one industry-specific forum. Use the founder’s real name, real photo, real credentials. Post substantive answers, not promotional ones. Six months of consistent presence is the minimum to be cited as an “expert perspective.”
4. Publish at least one extension piece per pillar this quarter. If you have a pillar page on, for example, “What is GEO,” publish a complementary deep-dive on a sub-topic — citation tracking methodology, a benchmarking dataset, or an industry-specific application. This feeds the “Further Exploration” slot.
5. Implement Article, Author and Organization schema on every page that does not yet have it. Hover previews and inline citations rely on the publishers Google can identify with confidence. Sites with Article schema and a verified Author Person entity are favoured. Sites with neither are guessed at.
What Does Not Change
The temptation after every Google update is to overcorrect. Three things still apply, and any GEO strategy that ignores them remains broken regardless of this week’s news.
Schema.org structured data still drives entity comprehension. Organization, LocalBusiness, Person and Service schema are the foundation. Without them, none of the new citation slots can identify you with confidence.
NAP consistency still drives Knowledge Graph anchoring. Name, address and phone number must match across your site, your Google Business Profile, your Wikidata entry, and your industry directory listings. Inconsistencies still suppress citation eligibility.
Cross-source verification still gates citation. AI models do not cite a brand based on what its own site claims. They cite based on what at least three independent sources confirm. Your About page does not earn citation. Your About page plus Crunchbase plus G2 plus Wikidata plus a journalism mention does.
The Bigger Picture: The Citation Economy Is Maturing
Google’s May 6 update is the clearest signal yet that AI search is no longer in beta. The product team is now optimising for retention, click distribution, publisher relations and trust labels — the same levers Google has historically used to govern the blue-link SERP. The citation economy is being formalised.
This has predictable consequences. The first is that citation slots will become more contested as more brands realise they exist. The second is that early movers will entrench. Once an AI model has cited a brand consistently for six months across thousands of queries, displacing that brand requires significantly more effort than acquiring the slot in the first place. The third is that measurement infrastructure will become non-negotiable. Tracking organic Google traffic in GA4 is no longer enough. Citation tracking tools — Otterly, Profound, Peec — and direct query monitoring across ChatGPT, Gemini and Perplexity move from optional to essential.
Australian brands have a small structural advantage right now. AI Overviews penetration in AU lags US by an estimated three to six months. The window to act on this update before it reaches saturation in your market is wider than the equivalent window in the US.
That window is not open indefinitely.
Does this update affect Australia, the US and LATAM equally?
No. Rollout is staged. The US receives most AI Search updates first, with AU and EN-LATAM markets typically reaching parity within six to twelve weeks. Spanish-language AI Mode coverage continues to lag English-language markets by an estimated two quarters.
When does the rollout complete?
Google did not provide a timeline for most of the new link features. Inline links and “Further Exploration” appear to be live in the US already. Expert perspectives and subscription labels are rolling out gradually.
Do I need to change my SEO strategy?
If your strategy already prioritises atomic content, schema, entity authority and topical depth, no — you are already aligned. If your strategy is keyword-targeted and volume-driven, yes. The CTR returns on rank-only optimisation continue to deteriorate.
Is FAQ schema still useful after this update?
Useful, yes. As of May 7, 2026, Google deprecated the FAQ rich result feature in standard search, but FAQPage schema remains parsed by AI Overviews, ChatGPT, Perplexity and voice assistants. We covered this in detail in our companion analysis on FAQ schema strategy.
What tools measure AI Overviews citation performance?
The current standard stack: Otterly.ai for ChatGPT and Perplexity tracking, Profound for cross-platform AI visibility, Peec.ai for citation analytics, and Google Search Console for AI Overviews appearance reporting (when available for your property).
How does Hyperdot audit citation patterns?
Our citation audit runs each commercial query for the client’s domain and top three competitors across ChatGPT, Gemini, Perplexity, Microsoft Copilot, and Google AI Mode. We record citation slot, citation context, source URL and supporting evidence. The output is a 90-day priority list of structural changes needed to capture missing slots.
Does this update favour large sites over niche sites?
Mixed. Inline links and atomic content reward well-structured small sites. Subscription labels favour publishers with paid products. Expert perspectives explicitly favour individual authors over brand-owned content, which gives small operators a path that did not exist before.
When will I see traffic impact in my analytics?
Citation appearance precedes traffic impact by four to eight weeks in our client data. Branded query volume in Search Console is the leading indicator. Direct traffic and citation referral traffic from chat.openai.com, perplexity.ai and gemini.google.com are the trailing indicators. The update arrived on a Wednesday morning. By Thursday, fewer than ten major SEO publications had covered the structural implications. By Monday, most agencies were still on the previous strategy. The brands that gain citation share over the next 90 days are the ones whose teams sat down with this update on the day it shipped, audited, and rebuilt. We do that audit for clients. We also do it for free for businesses that want to know where they stand before committing. Run your free AI Visibility Audit →
Author: Adriana Kligman, Founder and CMDO of Hyperdot. Google Partner. Meta Partner. Architect of the firm’s GEO methodology since 2016. LinkedIn
