Google’s AI Search Opt‑Out Is Here (In The UK). The Problem: You Still Can’t Measure What You’re Giving Up.

Google is finally offering a real opt-out from AI Search experiences—starting in the UK—and it’s a milestone publishers and businesses have been demanding for more than a year. But the rollout has a sharp edge: Google is also shipping AI performance reporting in Search Console that shows Impressions… and not much else.

If you’re a business owner or marketer, that’s not a small omission. It’s the difference between a decision you can justify to your CFO and a decision you make based on instinct, fear, or ideology.

In this editorial, I’ll break down what changed, what’s still missing, what could go wrong, and how to build an operating system for AI search visibility that doesn’t depend on wishful thinking. I’ll also share how we approach this at AYSA: monitor what’s happening, prepare improvements, ask for approval, and execute changes safely—because in AI search, “strategy” without execution is just commentary.

Concise summary

• In the UK, regulators have imposed obligations that push Google to give sites meaningful controls to withhold content from AI search features and AI training—without penalizing standard rankings.
• Google is testing a Search Console toggle that can exclude a domain from AI Overviews, AI Mode, and AI Overviews in Discover (domain-level only for now).
• Google is rolling out AI performance reporting in Search Console, but it currently shows impressions—not Clicks, Click-through rate (CTR), or clear separation of AI-feature traffic quality.
• That creates a business paradox: you have an “exit door,” but not the measurement you need to know whether walking through it helps or hurts.
• The right move for most SMEs is not immediate opt-out. It’s controlled experimentation, inventorying pages that are likely to be cannibalized, improving “citation-worthiness,” and building measurement proxies until click data becomes available.

Key takeaways (for busy operators)

• Opt-out is now a real lever. Treat it like a pricing decision or distribution decision—not a moral stance.
• Impressions alone are not performance. They’re a visibility signal. Helpful, but incomplete.
• AI search will change your funnel. More users will “learn” in the search results and convert later via brand or direct, which makes attribution harder.
• Prepare for a world of “earned citations.” SEO is expanding into AEO/GEO: Answer Engine Optimization / Generative Engine Optimization.
• Execution speed matters more than ever. The businesses that win will be the ones that can adjust content, structure, and internal linking weekly—not quarterly.

Table of contents

• What Changed This Week: Opt-Out Controls + AI Reporting (But Not The KPIs That Matter)
• How We Got Here: From “No Simple Way” To “Here’s The Toggle”
• What “Opt Out Of AI Search” Actually Means (And What It Doesn’t)
• The Measurement Gap: Impressions Without Clicks Is Not A Performance Report
• Why This Matters To SMEs (Not Just Publishers)
• A Practical Decision Framework: When To Stay In, When To Opt Out, When To Negotiate
• A Concrete SME Scenario: The Clinic That’s “Visible” But Not Growing
• What You Should Monitor Right Now (Even Without AI Click Data)
• What To Change On Your Site To Win In AI Search (Without Chasing Hype)
• What Agencies Must Rethink: Reporting, Retainers, And Accountability
• How AYSA.ai Helps: Monitor → Prepare → Ask For Approval → Execute
• What to do next (action list)
• Sources and further reading

What Changed This Week: Opt-Out Controls + AI Reporting (But Not The KPIs That Matter)

This week brought three overlapping developments, which together signal a new phase in AI search: regulators are now shaping product behavior, and Google is shipping controls faster than its measurement stack can support.

Based on reporting by Search Engine Journal, Google began testing a Search Console control that allows some site owners to exclude their domain from AI search surfaces—while keeping standard organic rankings intact. At the same time, Google started rolling out AI performance reporting in Search Console that breaks down AI feature visibility by page and country, but currently includes impressions only—not clicks or CTR.

The catalyst in the UK is regulatory: the Competition and Markets Authority (CMA) has imposed a conduct requirement on Google’s search business. The practical effect is that publishers should have a way to withhold content from AI search features and AI model training, and Google should not punish sites for doing so. The CMA also expects clearer attribution and reporting that allows publishers to evaluate outcomes.

Here’s the core tension: Google can say, “You’re free to opt out,” but if it doesn’t provide click-through and CTR data for AI features, businesses can’t quantify the cost of opting out (or the cost of staying in).

Research lead (secondary source): Search Engine Journal coverage of the rollout and the reporting gap: Google Gives Sites AI Search Opt-Out, But Not The Data To Use It.

How We Got Here: From “No Simple Way” To “Here’s The Toggle”

AI in search didn’t arrive with a single product launch. It arrived as a sequence of expansions:

• Google introduced generative summaries in search (AI Overviews) and then iterated on them.
• Websites, publishers, and SEOs pushed back: if Google uses your content to generate answers that reduce clicks, you should have meaningful controls and transparent reporting.
• Google provided partial controls (like robots directives that affect training) and partial visibility (more links in AI experiences), but not a clean “participate in AI answers, but don’t train models” vs “train models, but don’t show in AI answers” matrix that businesses can manage.

In the UK, the CMA’s work accelerated the timeline. As described in the SEJ reporting, the CMA designated Google with “strategic market status” in UK search and proceeded with conduct requirements. Google’s stance also evolved from “exploring” changes to “developing” them—language that matters when a regulator is watching.

Whether you’re in the UK or not, pay attention. The UK is becoming the test market where regulatory pressure and product controls are live at the same time. If it works there, it is likely to inform how similar conversations play out in other regions, including the EU and the US (the SEJ piece notes related policy contexts, but as an operator you don’t need to track every legal detail to understand the direction).

What “Opt Out Of AI Search” Actually Means (And What It Doesn’t)

When people hear “opt out of AI search,” they often imagine a single switch that keeps their content out of everything AI. Reality is messier, because Google’s AI touches multiple surfaces and functions.

Based on the SEJ reporting, Google’s Search Console toggle under test can exclude a domain from:

• AI Overviews (generative summaries in search)
• AI Mode (a more conversational AI search experience)
• AI Overviews in Discover (AI summaries in discovery feeds)

And importantly, Google has indicated that using the opt-out is not intended to be a ranking signal for standard search results. That’s a big deal: historically, “opting out” of snippet features or restricting snippet usage could indirectly affect visibility and clicks in classic organic. The whole point of this change is to separate “participate in AI features” from “compete in traditional organic rankings.”

But there are caveats:

• It’s domain-level (for now). That’s a blunt instrument. Many businesses have a mix of pages: some benefit from AI citations, others are cannibalized by AI answers. Page-level controls are the real requirement—and they’re not broadly available yet.
• It doesn’t automatically solve training vs display. Historically, directives like “Google-Extended” targeted training and grounding, while other tags (like “nosnippet”) affected display. The new toggle is about display in AI search surfaces. It doesn’t necessarily replace every other control you may be using.
• “Not a ranking signal” isn’t the same as “no consequences.” Even if rankings remain, your traffic mix can change. AI surfaces can suppress clicks on some queries while increasing qualified traffic on others (especially if citations become a trust signal).

This is why the missing measurement is so damaging. If your only feedback loop is impressions, you’ll naturally gravitate toward visibility theater rather than business performance.

The Measurement Gap: Impressions Without Clicks Is Not A Performance Report

Impressions tell you your content appeared in an AI feature. That is a form of “distribution,” but it’s not an outcome.

For a business, the KPI chain usually looks like:

• Visibility: impressions, share of presence, citation frequency
• Engagement: clicks, CTR, assisted visits, repeat searches for your brand
• Conversion: leads, purchases, calls, bookings, signups
• Unit economics: CAC, payback period, gross margin, LTV

Google’s new AI reporting in Search Console (as reported) currently gives you the first item—and not the second. But the second item is the hinge. Without clicks and CTR specific to AI features, you can’t answer basic operational questions:

• Are AI citations driving any traffic at all?
• If traffic is down, is AI the cause—or seasonality, competition, brand demand shifts, pricing, inventory, or product-market fit?
• Are AI-driven clicks higher intent than classic organic clicks?
• Which pages are “answer-complete” (users stop at the AI response) versus “action-needed” (users still click)?

The SEJ reporting also notes that the CMA’s interpretive notes expect Google to provide click-throughs, CTR, and AI-segmented data distinct from standard organic—delivered in an accessible platform like Search Console. That’s the right shape of requirement, because it makes the opt-out decision economically legible.

Until that arrives, any blanket conclusion (AI is good / AI is bad) is usually narrative masquerading as measurement.

Why This Matters To SMEs (Not Just Publishers)

It’s tempting to treat this as a “publisher problem.” But SMEs are increasingly publishers too—just with different monetization.

Consider the types of businesses that rely on informational content:

• Clinics publishing condition and treatment explainers
• Law firms publishing “what to do after…” guides
• Ecommerce brands publishing comparison pages and FAQs
• SaaS companies publishing setup guides and troubleshooting docs
• Local service companies publishing pricing and process explainers

For many of these, informational pages are the top-of-funnel engine that later converts through brand trust, email capture, or retargeting. If AI Overviews satisfy the informational need without a click, SMEs lose the first-touch session that feeds the rest of the funnel.

At the same time, AI citations can be a credibility amplifier. If your brand is cited as a source in the answer, you may get fewer clicks today but stronger brand preference tomorrow. That’s not a comforting answer to a revenue drop—but it’s a real strategic trade-off.

The problem is: without click/CTR reporting for AI surfaces, you can’t quantify the trade-off. So SMEs face two bad options:

• Opt out to protect traffic (maybe) and lose AI distribution (maybe valuable), or
• Stay in for exposure and accept traffic suppression without the ability to prove whether it’s “bounce clicks” or meaningful demand.

That’s why this moment matters: for the first time, we have controls and visibility signals showing up together. But we don’t have the decision-grade metrics yet.

A Practical Decision Framework: When To Stay In, When To Opt Out, When To Negotiate

Most businesses should not treat AI opt-out as a binary ideological choice. Treat it like you’d treat distribution on a marketplace, a referral partner, or a paid channel: you participate when it improves outcomes, you reduce exposure when it harms economics, and you negotiate when you have leverage.

Step 1: Inventory your “AI-exposed” query classes

Even without click data, you can categorize your content into buckets that behave differently in AI search:

• “Definition” queries: what is X, symptoms of Y, meaning of Z (high risk of zero-click)
• “How-to” queries: steps, checklists, troubleshooting (often answerable in AI; mixed click outcomes)
• “Comparison” queries: best X for Y, X vs Y (AI may summarize but users still shop around)
• “Transactional” queries: buy, book, pricing, near me (AI may assist but clicks often still needed)
• “Local intent” queries: location hours, services, insurance accepted (AI can misstate; accuracy is critical)

Step 2: Map each bucket to business economics

Ask:

• Which pages drive first-touch leads that convert later?
• Which pages drive direct conversions?
• Which pages exist mainly for brand trust (and could benefit from citations)?

If your content is heavily “definition” and monetized by ads or affiliate, the opt-out conversation is different than if your content is “pricing + booking” for a clinic.

Step 3: Assess risk tolerance and error cost

Sometimes the key factor isn’t clicks—it’s correctness.

If AI Overviews or AI Mode misrepresent your policy (returns, shipping, pricing tiers) or your medical/legal advice boundaries, the cost of being present could exceed the value of incremental exposure. In that case, opt-out (or at least page-level exclusion when available) becomes a brand safety decision, not an SEO decision.

Step 4: Run controlled experiments (where possible)

Today, experiments are limited because controls are domain-level for many. Still, you can:

• Segment by country (UK vs non-UK) where reporting is available and compare trends carefully.
• Track branded search demand and direct traffic alongside AI impressions to look for correlation (not proof, but signal).
• Use a subset property or subdomain strategy only if it’s already part of your architecture—don’t redesign your site around an experiment unless the upside is clear.

Step 5: Negotiate when you have leverage

The SEJ reporting notes that the CMA’s intent includes strengthening publishers’ negotiating position for content deals. SMEs may not negotiate directly with Google, but the principle still applies: if your content is uniquely valuable and you can credibly withdraw it, you can negotiate in other ways—affiliate partners, syndication, licensing, or even paid search allocations that offset organic uncertainty.

A Concrete SME Scenario: The Clinic That’s “Visible” But Not Growing

Let’s make this real.

Scenario: A multi-location physical therapy clinic invests in SEO content: “knee pain when climbing stairs,” “how long does plantar fasciitis last,” “what is dry needling,” plus location pages and service pages. Historically, these guides drive awareness and then patients book after reading, calling, or returning via branded search.

Now AI Overviews start answering symptom and timeline questions directly in the SERP. The clinic sees:

• Search Console impressions stay high (maybe higher)
• Organic clicks decline for informational pages
• Calls and bookings flatten
• Brand searches might rise slightly—or not, depending on market competition

With only AI impressions reporting, the clinic can’t answer:

• Are the AI citations sending any qualified visits?
• Are users seeing the brand in AI answers and converting later through brand search?
• Should the clinic opt out to protect top-of-funnel sessions—or stay in to remain the “recommended source”?

What I’d do in this scenario (without pretending we have perfect data):

• Protect the money pages. Ensure service pages and location pages are airtight: accurate, fast, conversion-focused, and clearly differentiated.
• Upgrade informational content to be citation-worthy. Add clear author/reviewer context, medically appropriate disclaimers, updated dates, and structured sections that AI can quote safely (more on this below).
• Increase the “next step” frictionless pathways. Put booking links, phone CTAs, insurance info, and “when to see a PT” decision points in the content—so if a click does happen, it converts.
• Measure assisted signals. Track brand search trends, call tracking, and landing page cohorts; treat AI impressions as distribution that may show up as assisted conversions later.

This is exactly where execution matters. It’s not enough to have an opinion about AI search. You need a weekly system that updates content, improves structure, and tests outcomes—without breaking the site or causing brand/compliance issues.

What You Should Monitor Right Now (Even Without AI Click Data)

If you’re waiting for Google to provide perfect AI click data before acting, you’ll be late. But you also shouldn’t guess blindly. Here’s a pragmatic monitoring stack that works today.

1) AI impressions by page and country (Search Console)

Use the new AI performance reporting to identify:

• Which pages show up most in AI features
• Which countries are affected first (especially if you operate internationally)
• Which content types are most “AI-summarized”

Even without clicks, this gives you an exposure map. Exposure maps tell you where to prioritize improvements and where risk concentrates.

2) Standard organic clicks and CTR (Search Console)

Watch classic organic performance for the same pages that show AI impressions. You’re looking for divergence patterns:

• AI impressions up, organic clicks down (possible cannibalization)
• AI impressions up, organic clicks stable (maybe AI is additive)
• AI impressions up, branded clicks up (possible assisted lift)

None of this proves causality, but it gives you a direction to investigate.

3) Brand demand and “second search” behavior (proxies)

When AI answers reduce first-click visits, your brand may still benefit through later behavior: users search your brand, revisit directly, or convert via other channels. That’s harder to attribute, but you can still monitor:

• Branded query impressions/clicks in Search Console
• Direct traffic and returning users in your analytics stack (e.g., GA4—if you use it)
• Call tracking, form submissions, bookings

I’m not going to pretend any single proxy perfectly replaces AI click data. The goal is triangulation: multiple imperfect signals that together reduce uncertainty.

4) SERP reality checks (manual sampling)

Pick 20–50 high-value queries and review them monthly (or weekly in volatile niches):

• Are you cited?
• Are competitors cited instead?
• Is the AI answer accurate in your category?
• Are there clear links out—or is the answer self-contained?

Document these. In a world where reporting is incomplete, disciplined sampling becomes part of the operating system.

What To Change On Your Site To Win In AI Search (Without Chasing Hype)

AI search rewards many of the same fundamentals as classic SEO—clarity, structure, and authority—but it changes the “unit of competition.” You’re no longer only competing for a blue link click; you’re competing to be cited, summarized, and trusted as a source.

Here are changes that are consistently rational regardless of what Google reports next.

1) Make your content easier to quote safely

AI systems gravitate toward content that is:

• Clearly structured (headings, short sections, scannable lists)
• Unambiguous (definitions, step-by-step instructions)
• Sourceable (facts tied to references, dates, and scope)

Practical upgrades:

• Add concise “answer blocks” at the top of key pages (without turning your page into a thin snippet farm).
• Use consistent definitions and terminology across your site.
• Include last updated dates and the criteria for recommendations (especially for comparisons).

2) Improve trust signals the way humans judge them

AI is trained on human language patterns, and it’s often aligned with human expectations for credibility. For SMEs, this is good news: you don’t need to be a media giant to be trustworthy.

• Show real authorship and expertise (bios, credentials where relevant).
• Publish clear policies (returns, shipping, privacy, editorial standards).
• Ensure contact details and business identity are consistent.

This aligns with the broader “E-E-A-T” mindset Google has discussed publicly in the past, though the specifics of how it’s used are not fully disclosed. The practical point: make your site obviously legitimate.

3) Strengthen internal linking for “AI journeys”

If AI reduces your informational clicks, the clicks you do get are more precious. Your site must convert them efficiently.

• Link informational pages to the next commercial step (service page, category page, booking page).
• Add contextual “related questions” sections that keep users engaged.
• Ensure core pages are reachable within a few clicks.

4) For local businesses: prevent AI from inventing details

Local is a high-risk category because wrong details cause real harm: wrong hours, wrong services, wrong insurance accepted, wrong pricing.

Practical steps:

• Keep location pages updated and consistent.
• Use structured content for hours, services, and FAQs.
• Reduce ambiguity: don’t bury critical details in images or PDFs.

5) Shift some content from “answer” to “decision”

If AI can answer “What is X?”, your opportunity is to own “What should I do next?”

Create content that includes:

• Decision frameworks
• Trade-offs and constraints
• Real examples and use cases
• Tools, calculators, templates (where appropriate)

AI can summarize these, but users still click when they need to decide, compare, or act. That’s where SMEs can outperform generic content farms.

What Agencies Must Rethink: Reporting, Retainers, And Accountability

For agencies, the uncomfortable truth is that AI search makes the old reporting model weaker.

Historically, agencies could show:

• Rankings
• Organic clicks
• Traffic growth
• Conversions attributed to organic

Now, even if rankings stay stable, AI can intercept user intent earlier. That means:

• Traffic might decline even when visibility is rising.
• Attribution shifts toward brand and direct.
• Clients feel like “SEO isn’t working” even when the brand is being used as a source.

Agencies need to add three capabilities:

1) AI visibility reporting (and education)

Clients must understand the difference between being ranked and being cited. If you’re cited but not clicked, that can still be value—if you can connect it to downstream outcomes.

2) AEO/GEO deliverables that are not “blog posts”

To compete in AI answers, agencies need to deliver:

• Content restructuring
• Entity consistency across the site
• FAQ design and information architecture
• Internal linking and conversion paths

Not just volume publishing.

3) Faster, safer execution cycles

The days of “strategy deck → wait two months → deploy” are over. AI search iterates quickly, and competitors will update pages constantly.

This is where execution systems become differentiators—especially ones that can propose changes and get explicit approval before pushing updates (critical for regulated industries and brand-sensitive teams).

How AYSA.ai Helps: Monitor → Prepare → Ask For Approval → Execute

At AYSA, our thesis is simple: modern SEO/AEO/GEO is an operational discipline, not a one-time project. The teams that win will be the ones that can measure what’s happening, decide what to change, and deploy improvements quickly—without chaos.

That’s why AYSA is built as an approved execution system:

• Monitor: Track AI search visibility signals and the pages likely impacted by AI answers. See more at AYSA Monitoring.
• Prepare: Identify what to improve: content structure, internal linking, technical cleanliness, and “citation readiness.” Start here: AI Search Visibility.
• Ask for approval: Every meaningful website change should be reviewed—especially for SMEs where the site is the business. That’s how you avoid well-intended changes that break conversions, compliance, or brand voice.
• Execute: Deploy accepted changes systematically, then monitor impact and iterate.

If you’re new to the space, our broader toolset is summarized at AYSA AI SEO Tools. For operators comparing approaches and budgets, see Pricing. And for ongoing thinking about search’s evolution, we publish additional editorials at AYSA Blog.

Why this matters in the opt-out era: the biggest risk isn’t choosing the “wrong” stance on AI participation. The biggest risk is being slow, unstructured, and unmeasured while your competitors adapt.

What to do next (action list)

Use this as a practical checklist for the next 30 days.

In the next 7 days

• Find your AI-exposed pages. Use Search Console AI reporting where available (or plan for it if not yet in your account).
• Classify those pages by intent: definition / how-to / comparison / transactional / local.
• Identify “business-critical” pages (money pages and reputation-sensitive pages) and confirm they are accurate, updated, and conversion-ready.

In the next 14 days

• Rewrite 5–10 high-exposure informational pages to be more decision-oriented, better structured, and more citation-ready.
• Add internal links from those pages to the next step (booking, category, product, contact).
• Document 20 priority queries and manually check whether you’re cited (and whether the answer is accurate).

In the next 30 days

• Build a simple AI visibility dashboard that pairs AI impressions with organic clicks, branded demand, and conversions. Even if imperfect, it keeps you grounded.
• Decide your provisional stance: stay in (default), opt out (brand safety), or wait for page-level controls (selective strategy).
• Prepare governance. If you operate in regulated categories, define who approves content changes, disclaimers, and page templates.

If you want help operationalizing this—monitoring, preparing prioritized changes, routing them for approval, then executing—start with AI Search Visibility and Monitoring.

Looking ahead: what to expect next (and what not to assume)

Three expectations are reasonable, given the direction described in the SEJ reporting:

• More controls will arrive. Page-level controls are a natural next step, and the CMA timeline suggests early 2027 for full implementation (per the SEJ piece).
• More reporting will arrive. Google has indicated it will add more data to AI reporting over time, but we should not assume the first additions will be the most business-useful unless regulators require it.
• Regional policy will shape product behavior. The UK rollout is likely to influence how similar requirements are discussed elsewhere.

What you should not assume:

• That AI impressions will translate into brand lift for your niche.
• That opting out will “restore” classic SEO performance—user behavior has already changed.
• That a single toggle is a strategy. It’s just a lever.

Sources and further reading

• Search Engine Journal: Google Gives Sites AI Search Opt-Out, But Not The Data To Use It (primary research input for this editorial)
• Search Engine Journal: Latest News (ongoing coverage of search changes)
• Search Engine Journal: SEO (context on SEO strategy shifts)
• AYSA.ai: AI Search Visibility
• AYSA.ai: AI SEO Tools
• AYSA.ai: Monitoring
• AYSA.ai: Blog
• AYSA.ai: Pricing

Note on sourcing: This article uses Search Engine Journal’s reporting as the research input. The SEJ piece references CMA interpretive expectations and Google Search Console testing details. Where official CMA documentation or Google product documentation is not included in the provided research context, I’ve avoided claiming specific legal text verbatim and focused on the operational implications described.

Related AI SEO resources

Continue the AI search topic inside AYSA.

Use these pages to connect the article with AI SEO tools, AI visibility monitoring, AI Overviews and approved website execution.

AI SEO tools The commercial hub for AI SEO, AI search optimization and approved execution. AI search monitoring Track AI visibility, answer-engine mentions and the next actions worth approving. AI Overviews SEO Prepare clearer, more structured pages without promising guaranteed inclusion. SEO, AEO and GEO visibility A plain-English explanation of the search visibility layer AYSA prepares for.

Written by

Marius Dosinescu

Marius Dosinescu is the founder of AYSA.ai, an entrepreneur focused on SEO automation, ecommerce growth, authority building and approved website execution for businesses that want organic growth without specialist overhead.

Author profile Dosinescu.ro LinkedIn

Share in X f @