Google Gives Sites AI Search Opt Out, but Not the Data to Use It

There is a specific kind of anxiety that comes with a single toggle switch in a dashboard. When that switch controls whether your content is used in AI search results, the stakes are not just about technical settings, they are about the future of your traffic. For a long time, publishers have felt like they were watching a black box, unable to tell if AI Overviews were stealing their visitors or acting as a new discovery mechanism.

Recently, Google introduced a way for some websites to opt out of these AI features without being penalized in standard search results. This happened alongside a new conduct requirement from the UK's Competition and Markets Authority, or CMA. On the surface, it looks like a win for publisher autonomy. But there is a catch that makes this a very difficult decision to make. While the door to exit is now open, the map telling us if we should actually walk through it is missing several key pages. This connects with structured data when the same signal needs a clearer operating decision. A useful companion note is UK Requires Opt Out, because it looks at a nearby part of the same system.

The core of the problem is a data gap. Google has started rolling out AI performance reports in Search Console, but these reports currently show impressions and not clicks. The CMA has explicitly stated that for a publisher to make an informed choice, they need to see click through rates and actual click data separated from organic search. Right now, that data is simply not there.

Expert Interpretation: The tradeoff here is between control and visibility. If you opt out now, you regain control over your intellectual property, but you do so blindly. The decision to flip a switch based on impressions alone is dangerous because impressions are a vanity metric in the context of AI. A high number of impressions in an AI Overview could mean your site is a primary source, or it could mean your content is being summarized so well that the user has no reason to ever click through to your site. Without click data, you cannot tell if the AI is a bridge or a wall.

The Regulatory Path to This Point

This situation did not happen in a vacuum. It is the result of a regulatory push in the UK. Back in October, the CMA designated Google as having strategic market status regarding search. This was a signal that Google's dominance required a higher level of scrutiny and specific conduct requirements to ensure fairness.

By January, the CMA opened a consultation on how Google should behave. Interestingly, Google's own language shifted during this process. In January, the company mentioned it was exploring updates to allow sites to opt out of generative AI features. By March, that language had evolved from exploring to developing. It suggests that the regulatory pressure from the CMA was the primary catalyst for the development of these controls.

Before this week, the options for publishers were fragmented and confusing. There was the Google Extended tag, which allowed sites to opt out of AI model training and grounding. However, using that tag did not guarantee that your content would stay out of AI Overviews or AI Mode. There was also the nosnippet tag, but that was a blunt instrument. If you used it to stop AI Overviews, you also lost your standard search snippets. You could not choose one without losing the other.

Expert Interpretation: This transition from exploring to developing shows that the industry is moving from a period of voluntary cooperation to one of mandated compliance. For publishers, this means the power dynamic is shifting. We are no longer just hoping Google provides a tool, we are seeing tools created because a government body demanded them. The lesson here is that regulatory pressure is often the only way to get transparency from platforms that benefit from keeping their data opaque.

Breaking Down the Recent Changes

Three distinct developments arrived this week, and it is important to distinguish between what is a legal obligation and what is a voluntary product update.

First, the CMA's conduct requirement is a legal mandate. It forces Google to allow publishers to withhold their content from both AI model training and AI search features. It also mandates that Google must clearly attribute domains in AI responses with links that lead back to the source. Crucially, the CMA has forbidden Google from penalizing any website that chooses to opt out.

Second, Google introduced a Search Console toggle. This is a voluntary product change that allows publishers to exclude their sites from AI Overviews, AI Mode, and AI Overviews within Discover. This happens at the domain level. Google has confirmed that this opt out will not be used as a ranking signal for standard search results. However, if you need more granular control, you will have to wait. Page level controls are not yet available, and the CMA has given Google until March 2027 to implement them.

Third, Google began rolling out AI performance reports in Search Console. These reports provide a view of how often your pages appeared in AI features, broken down by country and specific page. While Google says more data will be added over time, they have not provided a timeline or a list of what that data will be. The same pattern also shows up in Brand Signals Are Rewriting the Authority Stack, where the practical question is how the signal becomes visible.

Expert Interpretation: The most critical detail here is the promise of no ranking penalties. In the past, the fear of losing organic visibility has kept publishers compliant with platform changes they disliked. By removing that fear, the CMA has created a genuine choice. However, the gap between domain level controls and page level controls is a significant operational hurdle. Most large publishers do not want an all or nothing approach, they want to protect high value proprietary data while allowing AI to surface low value utility content. Until 2027, that nuance is unavailable.

The Critical Gap in Reporting

The tension lies in the difference between what Google has provided and what the CMA says is necessary. The CMA's interpretive notes outline three specific types of data that publishers need to make a rational decision about opting out.

The first is impressions, which simply shows when content appears in AI features. Google has delivered on this. The second is engagement data. This includes actual click throughs from AI features to the publisher's website and a way for publishers to identify those clicks to assess their quality. The third is the click through rate, or CTR, which is the percentage of users who actually click the link within the AI feature.

The CMA also specified that this data should be separated from organic search results and delivered through a common platform like Search Console. Currently, the reports only cover impressions. The clicks and the CTR are missing. This leaves publishers in a position where they can see that they are being used, but they cannot see if they are being rewarded with traffic.

Expert Interpretation: This is where the implementation feels incomplete. Impressions without clicks are essentially a measure of how much of your content is being consumed within Google's ecosystem without the user ever leaving. If your impressions are skyrocketing but your clicks are flat or falling, the AI is effectively cannibalizing your traffic. By providing only impressions, Google is showing us the reach but hiding the impact. Any publisher making a decision based on the current reports is essentially guessing.

The Strategic Risk of the Exit Door

Freelance SEO consultant Natalie Arney put this perfectly on LinkedIn, noting that one announcement gives publishers the exit door, while the other shows what it would cost to walk through it. This is the central dilemma for every site owner right now.

If a publisher opts out now, they might be walking away from a significant stream of traffic that they cannot yet measure. They are protecting their content from being used to train models, but they are potentially cutting off a new discovery channel. On the other hand, if a publisher stays in, they are operating on a baseline of impressions. They know they are visible, but they do not know if that visibility is beneficial or parasitic.

For those advising clients, these AI performance reports are the first dedicated view we have had of how a site appears in AI responses. Even if the data is incomplete, it provides a baseline that did not exist before. It allows us to see which pages are being targeted by AI, which gives us a clue as to what Google considers high value or easily summarizable content.

Expert Interpretation: The decision to opt out should be viewed as a risk management exercise. If your business model relies on high volume, low intent traffic, the risk of opting out is higher. If your business model relies on deep authority and proprietary data that you would rather sell or gate, the risk of staying in is higher. The lack of click data means you cannot use a data driven approach, so you must instead use a business model approach to decide your stance.

What to Watch for Next

The timeline for these changes is staggered. The conduct rule is in effect immediately, but other obligations do not start until December. The long lead time for page level controls, stretching into early 2027, suggests that this will be a slow rollout.

The CMA has indicated that they will announce further actions regarding Google's search business in the coming weeks. The most important thing to watch for is whether the reporting catches up to the controls. If Google adds click throughs and CTR to the Search Console reports soon, the opt out tool becomes a precision instrument. If the reporting remains limited to impressions, the tool remains a gamble.

The measurement gap is the real story here. Until publishers can see the actual flow of users from AI Overviews to their own domains, the ability to opt out is a gesture of autonomy that lacks the necessary information to be practical. We have the switch, but we are still waiting for the light to turn on.

Introduction

The key issue here is Some websites can now opt out of Google's AI search features without losing their place in standard search results. The UK's Competition and Markets Authority imposed a conduct requirement this week, and Google began testing its own Search Console toggle the. My read is to treat it as a decision point: what signal needs to become clearer, what part of the system is currently weak, and what evidence would show that the work is improving visibility rather than only adding activity.

That is the difference between reacting to a trend and building a useful search system. Connect this point back to the page template, internal linking, entity signals, content depth, crawl accessibility, and the way the brand is represented across the wider web before deciding what to change first.

Why This Matters

The key issue here is Freelance SEO consultant Natalie Arney connected both announcements on LinkedIn: "One gives publishers the exit door. The other shows what it would cost to walk through it." That's the decision publishers face now. The opt out exists, but the data to. My read is to treat it as a decision point: what signal needs to become clearer, what part of the system is currently weak, and what evidence would show that the work is improving visibility rather than only adding activity.

That is the difference between reacting to a trend and building a useful search system. Connect this point back to the page template, internal linking, entity signals, content depth, crawl accessibility, and the way the brand is represented across the wider web before deciding what to change first.

Looking Ahead

The key issue here is The conduct rule takes effect immediately, while other obligations start in December. The nine month implementation for page controls points to early 2027. The CMA will announce further action on Google's search business in the coming weeks. Google's reports. My read is to treat it as a decision point: what signal needs to become clearer, what part of the system is currently weak, and what evidence would show that the work is improving visibility rather than only adding activity.

That is the difference between reacting to a trend and building a useful search system. Connect this point back to the page template, internal linking, entity signals, content depth, crawl accessibility, and the way the brand is represented across the wider web before deciding what to change first.

Practical next steps

The useful part is not only the idea itself, but the operating habit behind it. Use it as a checklist for decisions: what deserves attention now, what should be monitored, what needs a stronger evidence base, and what can wait until the system has more scale.