Google Just Killed Search As We Know It

For twenty-five years, Google operated on a simple deal. You search, they show links, you click one, some website gets a visitor. The entire digital economy — publishing, ecommerce, performance marketing, that blog your cousin started about artisanal candles — was built on this exchange continuing forever.

At I/O 2026 on Tuesday, Google made it clear that deal is off.

From Librarian to "I'll Just Do It Myself"

Here's what Google actually announced, and it's worth laying out because each piece alone would be a big deal. Together, they describe a search engine that no longer wants to send you anywhere.

Generative UI lets Search build interactive tools — dashboards, comparison widgets, simulations — directly inside the results page. The demo showed a student asking about black holes and getting a manipulable physics model right there in the browser. No click required. No website visited. Just Google going, "Why would you need anyone else?"

Mini apps let users build persistent tools — meal planners, fitness trackers, wedding dashboards — using natural language prompts, all without leaving Search. (Because apparently Google looked at the entire app economy and thought, "We could just absorb that.")

Information agents run continuously in the background, scanning blogs, news sites, and social media, then synthesising updates and delivering them proactively. They don't surface content for you to read. They read it for you. Your carefully crafted product comparison article? The agent already digested it, extracted the useful bits, and served a summary to the user who will never know your site exists.

Generative UI rolls out free this summer. Mini apps and information agents launch for Google AI Pro and Ultra subscribers shortly after. AI Mode already reaches over one billion monthly users. AI Overviews touch 2.5 billion.

Read that last paragraph again. This isn't a beta test for enthusiasts.

The Traffic Picture Was Already Ugly. It Just Got Worse.

The data heading into I/O was grim enough. Roughly 58% of Google searches already end without a single click to a third-party website. Ahrefs research found AI Overviews correlated with a 58% reduction in click-through rates for top-ranking pages — position-one CTR dropping from 7.3% to 1.6% on affected queries. Seer Interactive found organic CTR fell 61%. Chartbeat data from over 2,500 news sites showed Google search referrals declined 33% in 2025.

Google knows this is a problem. Before I/O, they rolled out five changes to improve link visibility — suggested angles linking to deeper articles, subscription labels, firsthand perspectives from Reddit, inline links within AI responses, and hover previews on desktop.

But here's the thing: those fixes were designed for a world where Search still answers questions with text and offers links alongside the answer. Tuesday's announcements describe something fundamentally different. Generative UI doesn't answer your question and offer a link. It builds you a tool. Information agents don't surface content for you to read. They read it for you. The distance between "AI Overview with better link placement" and "AI builds you a custom dashboard" isn't a product iteration. It's a category change.

The Brand Divide Is Now Measurable (And It's Brutal)

Here's the data point that matters most for anyone selling things online, and it was important even before Tuesday sharpened its implications.

Research from Amsive found that branded queries — searches where users type a specific brand name — with AI Overviews see an 18% increase in click-through rate. Generic queries, by contrast, see a 34% to 46% decrease. Seer Interactive's study showed that brands cited within AI Overviews earned 35% more organic clicks and 91% more paid clicks compared to brands that weren't cited.

The pattern is brutally clear. If people already know your name and search for it directly, AI-mediated search actually helps you. If your visibility depends on ranking for generic category terms and hoping someone clicks through — the economics are moving decisively against you.

Think about what this means for a mid-sized Amazon seller who's spent years optimising for "wireless noise-cancelling headphones under £100." That generic keyword strategy was already expensive. Now Google is building comparison tools directly in the search results, and the only brands showing up inside those tools are the ones the AI already considers authoritative — or the ones the shopper had in mind before they started typing.

The middle ground — competent but undifferentiated brands relying on keyword positioning to capture generic intent — is being compressed from both sides. Brand recognition and AI citation authority on one side, and Google's own shopping infrastructure eating the discovery layer on the other.

SEO Is Becoming a Citation Strategy (Whether You Like It or Not)

Traditional SEO was built on a straightforward model: create content that matches search intent, earn a high ranking, convert the resulting traffic. Every step of that model is now under pressure.

The shift isn't from SEO to something completely different. It's from SEO as a traffic acquisition strategy to SEO as a citation strategy. Being referenced within an AI-generated experience — "According to [Your Brand]" in an AI Overview, or having your product data surfaced when an agent builds a comparison — delivers value even without a click.

Google evaluates how consistently a topic is covered across a website, meaning depth and genuine expertise matter more than any single page optimised for a target keyword. The brands producing original research, distinctive category analysis, and comprehensive product information are structurally advantaged. Generic buying guides — the kind your listing specialist can produce in an afternoon — are being devalued at an accelerating rate.

For Amazon sellers specifically: Google's new Universal Cart aggregates products across retailers into a single AI-powered shopping cart with price tracking and compatibility checking. If a shopper asks Google's AI to help them build a custom PC, the agent pulls parts from multiple retailers, surfaces price history, and flags compatibility issues. That entire discovery process bypasses Amazon search. The sellers who benefit are those whose product data is comprehensive enough for Google's systems to interpret accurately — which brings us to the feed infrastructure question.

Your Product Feed Just Became Your Most Important Strategic Asset

Buried beneath the headline announcements, Google Merchant Center quietly rolled out a beta: "Use AI to add products." The tool performs a one-time AI scan of a merchant's website and populates product listings automatically. Point Google at your site and it figures out what you sell. Lovely.

The catch: it's a one-time scan. No continuous sync. Pricing changes, stock updates, and new products won't propagate after the initial scan. For anyone in categories where pricing shifts daily or inventory turns over quickly, the data degrades from the moment the scan completes.

Here's the deeper tension. Product feed quality is no longer just an advertising input — it's the foundation of visibility across an expanding set of surfaces. Merchant Center data now powers Shopping ads, free listings, AI Mode results, the Gemini shopping experience, Universal Cart, virtual try-on in Google Lens, and connected TV formats. The same feed data that AI Max reads to generate your ad copy determines how your products appear when an AI agent evaluates options on a shopper's behalf.

An AI-scanned catalogue is unlikely to match the attribute depth of a carefully maintained feed. The merchants who've invested in structured, comprehensive product data — and endured the tedium of cleaning feeds, resolving errors, and maintaining compliance — hold a structural advantage that a one-time scan cannot replicate.

The scan tool is a starting point. It is emphatically not a strategy.

Google's Advertising Black Box Just Got Bigger

The I/O announcements didn't happen in isolation. In the days before the keynote, Google released a cluster of advertising updates that complete the picture.

AI Max, Google's fastest-growing AI Search ads product, is extending to Shopping campaigns. The system reads your Merchant Center feed attributes, generates ad copy dynamically, and serves those ads across AI Overviews and AI Mode. Google's AI reads your product feed, writes your copy, and decides where and when to show it — including inside conversational search experiences where format and user intent differ substantially from traditional results pages.

A new "AI Brief" feature lets advertisers guide this in natural language — "include pricing" or "don't show ads for queries including 'inexpensive.'" This is Google's answer to the black box complaint. But rather than restoring granular targeting controls, Google is offering a conversational interface where you express preferences and the AI interprets them at its discretion. You describe what you want. Google decides how to deliver it.

The performance numbers are real. Smart Bidding Exploration delivers 27% more unique converting users. Demand-led pacing cuts manual budget adjustments by 66%. But the algorithm is finding queries you didn't target, bidding at prices you didn't set, and generating conversions you can't fully attribute to specific decisions.

For PPC managers running Amazon and Google Shopping simultaneously, the reality is that each platform is independently making more decisions on your behalf, using signals you can't independently verify. The advertiser's role is shifting from "operator" to "auditor of an AI's homework."

The Sceptic's Case Still Matters

Not everyone at I/O was convinced. Engadget's reporter questioned whether anyone is actually asking for agentic coding inside a search engine. Fair point. Google has form on announcing ambitious I/O features that ship late, ship limited, or quietly vanish.

Google Photos pushed an AI-only search overhaul and reversed course after user backlash. The AI Overviews track record (suggesting people eat rocks, put glue on pizza) is a useful reminder that generating custom interfaces billions of times daily without hallucinations is a substantially harder problem than doing it in a controlled demo.

The strategic direction, however, is unambiguous — even if the execution stumbles.

The Bottom Line

Google didn't update Search on Tuesday. It declared its intention to replace what Search means. The transition will be messier than the keynote implied, and some features may not survive contact with actual users. But the direction is set.

The brands that invested in being known, trusted, and sought out by name — the ones whose product data is comprehensive enough that AI systems cite them as authorities rather than bypassing them for a competitor — are the ones best positioned for what comes next.

For everyone else, the runway just got meaningfully shorter.

P.S. If your current SEO strategy is "rank for generic keywords and hope for clicks," this is your formal notification that the hope part of that equation is doing increasingly heavy lifting.

P.P.S. Google literally announced a feature where Search builds you a custom app instead of showing you a website. If that doesn't make you rethink your product data investment, I'm not sure what will.

P.P.S. On the bright side, at least we don't have to worry about "I'm Feeling Lucky" anymore. Google replaced it with an AI Mode button. Which, when you think about it, is the most honest rebrand in tech history — because nobody was feeling lucky with those AI Overviews anyway.

• OpenAI launched a standalone deployment company backed by $4B and 19 firms including McKinsey, Bain and Goldman Sachs, to embed engineers directly into enterprise operations.

• 40% of B2B sites have no homepage schema. Run your own company through an incognito Claude chat to see exactly what LLMs can and cannot read about you.

• Amazon launched Dynamic TV Creative on Prime Video, personalising interactive ads using shopping and browsing signals with no extra input needed from advertisers.

📖 Reconfigured - The analysts' journal for recording findings.

⚙️ Agent Minimax - Research, PPT, Code and more with Minimax . A global AI foundation model company.

✍️ SEO.ing - Your Go-To Solution for Quick, Effective, and SEO-Optimized Article Writing

💬 CopilotKit - Build in-app AI chatbots and AI-powered text areas

💻 Brainner - Streamline talent acquisition by automating AI-driven resume analysis

😍It was great, all I needed

🙂Good, not great

🙄It sucked