Hi everyone!

The news this week is dense, but one question ties it all together: who gets picked by AI answers? Google announced Preferred Sources is now inside AI Overviews and AI Mode, with the count hitting 345K. Marie Haynes installed the widget on her own site in 15 minutes. Peec.ai's 500,000-prompt study showed AI Overviews now appear on 87% of searches. Gianluca Fiorelli mapped how spam is mutating inside AI search. Google released its official AI optimization guide. Sundar Pichai told Decoder that "search, agents and tools will become one product." OpenAI is building conversion-focused ad infrastructure inside ChatGPT.

Grab your coffee, let's dig in.

Google announced on May 27, 2026 that Preferred Sources is now part of AI Overviews and AI Mode. Users have so far selected 345,000 sources — roughly four times the 90,000 count from December 2025. Preferred Sources is no longer limited to Top Stories; the badge now appears on links inside AI-generated answers. According to Google, users click on Preferred Sources at twice the rate of other links. This validates a position I've held for years: in a zero-click world, your audience itself is the most reliable visibility lever. If I had to give you one action this week, it's this — install the widget on your site today.

There are actually three updates bundled in this release. First, the Preferred Sources badge now shows up on links inside AI Overviews and AI Mode responses. Search Product Manager Duncan Osborn says users click on Preferred Sources at twice the rate of other links. Google didn't share the methodology behind that metric, so I'd read the "2x" number with some caution — but the direction is clear. Second, new article and "Perspectives" carousels are rolling out for developing topics. These will highlight both timely news links and first-hand perspectives sourced from forums and social media. Third, the "Highly Cited" label is expanding into standard search results, with a new second label flagging when an article "explicitly references a Highly Cited source." That means content that openly attributes original reporting gets a visibility bump in the AI lens too.

Marie Haynes shared in her May 27 video that she installed the widget on her own WordPress site in roughly 15 minutes. She gave Antigravity a natural-language prompt — "build me a widget for Google's new Preferred Sources feature" — and the agent read the docs, generated an HTML snippet for the Code Snippets plugin, and even matched the styling to her brand colors. Every blog post now ends with an "Add mariehaynes.com as a Google Preferred Source" link. The format itself is dead simple: just append your URL to https://google.com/preferences/source?q=. Haynes' best note: "This is an opportunity for anyone with brand followers to see themselves more often in AI search. Followers support the author they like, and they enrich their own answer feeds at the same time."

Peec.ai's Tomek Rudzki published a 500,000-prompt study on May 28 with two striking data points. First, AI Overviews now appear on 87% of searches; that's up from 56.9% in April 2025, a 50% jump in one year. Second, and more importantly, AI Overviews are not just hitting informational queries — they show up on 88.5% of decision-stage (bottom of funnel) queries. So when someone types "best CRM for small business" or "Salesforce vs HubSpot," Google's AI is delivering purchase advice to 2.5 billion monthly users. If your brand is not inside that advice, you are invisible — and the more invisible you are, the more the Preferred Sources badge multiplies its value, because once a user has flagged you as preferred, you appear more easily in their next query too.

Here's a roadmap that won't eat hours of your day. I wrote it assuming WordPress, but the logic translates to any CMS.

1) Prepare the target URL. Format: https://google.com/preferences/source?q=stradiji.com. Drop your domain into the q parameter, nothing else.

2) Write the widget copy. Open with something soft like "If this content helped you" and follow with "Add Stradiji.com as a Preferred Source on Google." Avoid marketing language. Explain plainly what you're asking and why: "If you'd like to see me more often in Google's AI answers, click this link and add Stradiji.com to your Preferred Sources." That's enough.

3) Place it. End of every blog post, not category archives. Add it to high-intent pages like your homepage and about page too. It's a 1-2 minute job per template; plugins like Code Snippets or Insert Headers and Footers let you push it sitewide in 5 minutes without touching theme files.

4) Tell your audience. Send a single post to your email list, LinkedIn, and X with a "Add me as a Google Preferred Source" call. Example: "If you'd like to see SEOs Diners Club more often in Google's AI answers, click here and add Stradiji.com to your Preferred Sources. One click, five seconds." Don't overcomplicate the language; say what you want.

5) Measure it yourself. Fire a custom event in GA4 called add_preferred_source_click and track click counts. Google does not report this metric back to individual sites, so build your own measurement loop and track how much your audience is converting.

6) Build the cadence. Drop a short reminder in your newsletter once a month. New readers keep arriving; old readers have already added you. Keep the reminder light and never pushy.

Before you flip the switch, sit with these three questions. First, who is loyal to you? Your email list, podcast listeners, YouTube subscribers, community members, client portfolio — these are the natural audience for a Preferred Sources widget. Cold traffic does not engage with this mechanic. Second, what topics do you actually own? A site flagged as a Preferred Source surfaces more often in queries about that topic. If you're an authority in a narrow niche, the badge effect compounds. Third, how fresh is your publishing cadence? Google specifically called out "fresh content" in the announcement; a blog that publishes twice a year won't extract full value from the badge. The coming weeks are a good time to tighten your publishing calendar.

Google rolled out the May 2026 Core Update on May 21, and we're in the final stretch of the two-week rollout. This is 2026's second core update and the fourth confirmed ranking update of the year. Notably, Google didn't publish a blog post or state any goals this time.

My field observation is clear: generic, synthesis-style content is bleeding out, while content built on first-hand data and original expert analysis is winning. Try a practical test: hand a piece of your content to ChatGPT, Claude, or Gemini and say "produce a synthetic copy of this." If the model can generate something similar in minutes, your content is not "unique" through the lens of this update. I've been running this test routinely in client content audits for the past three months, and roughly 60% of the blog posts I review fail it. As an expert, I'd boil 2026's new content standard down to one sentence: "Write content that can't exist without your field experience."

The intersection with AI Mode is critical here. Queries in AI Mode are three times longer than traditional search, and follow-up queries in the U.S. are growing 40% month-over-month. A page that weakens in the core update both slides on blue links and drops out of AI Mode's selection pool. The decline isn't single-layer — it's compounded.

Gianluca Fiorelli's May 27 post mapped how classic Google spam tactics are evolving for AI search. I'm summarizing this here because it gives perfect context to the question "why are we leaning into explicit signals like Preferred Sources?" Cloaking has become "User-Agent Conditional Prompt Injection" — sites serve one version to AI crawlers and another to humans. Doorway pages became "RAG Ingestion Flooding" — spam bots flood thousands of fake forums and Reddit threads to poison the AI's vector database. Comment spam evolved into "RAG Sentiment Overloading" — LLM bots conversationally praise specific brands across Reddit and Quora at scale. Link buying gave way to "LLM Training Dataset Placement" — paying high-authority archives to embed brand names into the model's base memory.

Fiorelli's core warning: AI search engines aren't ready for these attacks yet. Google, OpenAI, Anthropic, and Perplexity each have different blind spots in the comparison. This doesn't reduce the value of organic GEO work; it amplifies it. Two takeaways from me. First, user-controlled signals like Preferred Sources are the most manipulation-resistant visibility lever you have. Second, getting close to spam tactics is serious risk: Google's May 2024 expired domain abuse policy is now enforced more aggressively in the AI context. My advice to clients is simple: there are no shortcuts — audience + first-hand data + clean technical foundation is the triangle that wins.

Google released its "Optimization Guide for Generative AI Features in Google Search" and it's worth reading carefully. Two angles matter. First, Google formally defines "AEO" and "GEO" and then immediately writes: "Optimizing for generative AI search is really just optimizing the search experience, and that's still SEO." This is exactly the line Mike King attacked in his May 18 iPullRank piece.

Second, the guide lists four things "you don't need to do": LLMs.txt files are unnecessary, content chunking is unnecessary, writing specifically for AI is unnecessary, and over-focus on schema markup is unnecessary. Watch the nuance: Google says these are unnecessary for Google Search. Marie Haynes' May 29 analysis of the Lighthouse "agent-readiness" report reaches the opposite conclusion on LLMs.txt — not valuable for search, but valuable for the agentic web.

This actually reinforces a position I've held for a long time: a GEO roadmap can't be built for Google alone. ChatGPT, Perplexity, Claude, Copilot, and Gemini weigh different signals differently. A tactic Google calls "unnecessary" can be meaningful for non-Google engines. The benefit of the official guide is that now, when an enterprise client asks "what does Google actually say about this?", you can point to the source document directly. Just don't build a strategy on a single source.

In her May 29 post, Marie Haynes walked through Google's new Lighthouse "agentic browsing" report step by step. The report runs inside Chrome Canary's Developer Tools right now. It doesn't give you a 0-100 score; it shows the ratio of "agent-readiness" checks your page passes. There are four main categories: AI Accessibility Tree, WebMCP integration, Cumulative Layout Shift, and LLMs.txt format.

Three practical takeaways. First, accessibility is now dual-use. The accessibility tree, originally designed for screen readers, has become the map agents use to read your page. ARIA roles, semantic HTML, clear labeling. My advice to clients: run a simple accessibility audit this week and save the report. Second, pay attention to WebMCP. Declarative WebMCP means wrapping code around a form or tool so the agent sees it as a "structured tool." Imperative WebMCP goes further, enabling two-way conversation between agents and your site. Last month at Search Central, Google said "don't worry about this yet" — this Lighthouse release changes that signal. Third, LLMs.txt is valuable for agentic use cases. It's not required if you're a pure content site, but it's valuable on sites with functional tools (forms, scheduling, calculators) that agents will operate.

My take is clear: over the next 12 months, "agent-readiness" will become a new layer of technical SEO. Run a simple Lighthouse "agentic browsing" audit this week, then come back to it in three months.

Penske Media's federal court filing in the antitrust case it opened in February 2026 became the subject of detailed analyses this week. According to evidence Penske submitted, Google's AI summaries are reducing publisher clicks by 58%. Ahrefs' February 2026 research found a 58% click-through drop on top-ranking pages — nearly double the 34.5% decline from April 2025. Pew Research's study of 68,000 real queries found users click only 8% of the time when AI Overviews appear, versus 15% when they don't. Treating the traffic decline passively is no longer an option; Preferred Sources, growing branded search, and parallel channels like email + YouTube are the new three-legged foundation.

In his May 26 Decoder interview, Sundar Pichai summed up Google's mid-term vision in a single sentence: "Search, agents, and tools will become one product." This matters because in an environment where AI Mode and AI Overviews are being merged and Project Mariner is testing browser agents, the direction is settled — users won't type queries, agents will browse the web for them. SEO takeaway: explicit signals like agent-readiness and Preferred Sources shape the answer to "where will the agent cite me from?"

According to Search Engine Land's May 26 report, OpenAI is building conversion-focused ad infrastructure inside ChatGPT. A performance-billed format aimed at local businesses is being tested: the advertiser pays when a user books an appointment, makes a purchase, or fills out a form. You'd install OpenAI's own pixel on your site and feed conversion data back through the API. This puts ChatGPT in direct competition with Google Ads and Meta Ads. For ecommerce and local service clients, this format is the agenda for the next six months.

YouTube announced on May 27, 2026 that it's moving AI content labels to more visible spots and adding automatic detection. On long-form videos, the "AI" label moves out of the description and right below the player; on Shorts, it overlays the video itself. YouTube says there's no algorithmic penalty, but viewer behavior generates a secondary signal. Lean into human-made, on-the-ground footage in your channel identity; for the answer segments Ask YouTube clips and reuses, this is a double win.

Anthropic released Claude Opus 4.8 on May 28, 2026. One-million-token context window by default, 2.5x faster responses in fast mode, no price change. A Mythos-class model is expected in the coming weeks. Perplexity Comet has opened to enterprise, integrated with Samsung Internet as a default search option. Comet is an "agentic browser" — the agent does the browsing. Structured data, clear answer blocks, and BLUF format are critical not just for search engines now, but for the browser reading the site on the agent's behalf.

Mark Williams-Cook argued in his May 27 piece that the evidence behind claims that "schema markup increases citations in LLMs" is alarmingly thin. This aligns with the "don't over-focus on schema" line in Google's official guide. My read: schema still has value, but the "if I add this I get cited" framing is wrong. Schema is a supporting structure for original content — not a shortcut.

Chris Long wrote on X on May 27 that freshness and original data signals are now the new priority for AI Overviews source selection. Same direction as Liz Reid's "richer and deeper content" comment, and consistent with what we're seeing on the ground in the May 2026 Core Update.

Preferred Sources now lives inside AI Mode as a badge — installing the widget on your site today is the clearest lever for turning your audience into AI visibility in a zero-click world, and over the next 12 months "agent-readiness" will become the technical layer that completes it.

None of the topics this week work alone. Installing the Preferred Sources widget, reading the Lighthouse agent-readiness report, rebuilding pages weakened by the core update — these fit together when you work through them in sequence. If you'd like to keep up with these moves in real time, SEOs Diners Club is where I share the weekly read; if you want to go deeper with strategy on specific markets or verticals, that's what Stradiji does for enterprise clients in the US, UK, and Australia.

See you next week with more.

Mert Erkal

I genuinely enjoy sharing what's moving in SEO and AI search with you every week. If you find the work useful, you can keep it going by buying me a coffee. ☕ Click 'Buy Me a Coffee' to contribute to the knowledge-sharing. Let's build more together.

Mert Erkal is a digital marketing and SEO expert with 15+ years of experience. As the founder of Stradiji, he provides consulting on SEO strategy, conversion rate optimization, and AI integration to global brands across the US, UK, Australia, and beyond.

More information: