Google AI Overviews Are Stealing Clicks. Here Is What the Data Shows

Google AI Overviews are reducing organic click-through rates on informational queries, and the scale of that reduction is larger than most SEO teams have accounted for. When Google answers a question directly at the top of the page, a meaningful share of users never scroll further. That is not a prediction. It is what the traffic data from 2024 and early 2025 is showing across multiple industries.

The question worth asking is not whether AI Overviews affect SEO. They do. The question is which queries, which content types, and which sites are most exposed, and what a commercially sensible response looks like.

What the Traffic Data Is Actually Showing

The honest answer is that the data picture is still forming, and anyone claiming certainty about precise percentage drops should be treated with scepticism. What is clear is the directional trend. Semrush published analysis in 2025 showing that queries triggering AI Overviews correlate with lower click-through rates compared to equivalent queries without them. Their AI search and SEO traffic study is worth reading in full because it breaks down the impact by query type rather than treating all organic traffic as a single category, which is the only analysis that actually helps you make decisions.

The pattern that emerges is predictable once you think about it. Informational queries, the kind that begin with “what is”, “how does”, “why does”, see the heaviest AI Overview coverage. These are also the queries where a two-sentence answer genuinely satisfies the user’s need. Google has identified that, and it is serving that answer without requiring a click. Commercial and transactional queries, where the user needs to compare, buy, or book, are less affected because a generated summary cannot replace the actual transaction.

I spent years at agencies where a significant portion of SEO investment was directed at informational content on the assumption that top-of-funnel visibility would eventually convert. That assumption was always shakier than people admitted, because attribution between an informational blog post and a purchase three weeks later was rarely clean. AI Overviews have not created a new problem so much as they have made an existing one more visible and more urgent.

If you want a broader view of where SEO strategy is heading in light of these changes, the Complete SEO Strategy hub covers the full picture, from keyword architecture to content investment decisions to technical foundations.

Which Content Types Are Most Exposed

Not all content is equally at risk, and treating AI Overviews as a uniform threat leads to the wrong response. The exposure varies significantly by intent type, content format, and how much the query requires something beyond a factual answer.

Definitional content is the most exposed category. If your site has pages built around “what is [term]” queries, those pages are competing directly with Google’s ability to synthesise a clean definition from multiple sources. The AI Overview wins that contest most of the time, and it will win it more often as the technology improves.

How-to content sits in a more complicated position. Simple procedural content, “how to change a password”, “how to format a spreadsheet cell”, is vulnerable for the same reason definitional content is. The answer is short, factual, and easily synthesised. But complex how-to content, the kind that involves judgement, trade-offs, sequencing decisions, or professional context, is harder for AI Overviews to handle well. The generated summaries tend to flatten nuance, which creates an opportunity for content that explicitly addresses the complexity rather than pretending it does not exist.

Opinion and analysis content is the least exposed category, and this is worth sitting with for a moment. AI Overviews are built on synthesis. They aggregate existing positions. They cannot generate a genuine perspective, a point of view grounded in specific experience, a counterintuitive claim backed by reasoning that has not appeared elsewhere. That is not a capability gap that will close quickly, because it is not a technical problem. It is a structural one. Synthesis cannot produce original thought.

I judged the Effie Awards a number of years ago, and one thing that experience reinforced was how rarely the winning work looked like a synthesis of existing best practice. The campaigns that drove measurable business outcomes tended to have a specific, often uncomfortable, insight at their core. The same principle applies to content. The pieces that hold their position in a world of AI-generated summaries will be the ones built around a perspective that cannot be synthesised from what already exists.

What Being Cited in an AI Overview Is Worth

There is a strand of optimism in the SEO community that being cited as a source within an AI Overview compensates for the click-through rate reduction. The logic is that if Google cites your content, you receive a credibility signal and some referral traffic. Both of those things are partially true and both are overstated.

The referral traffic from AI Overview citations is real but inconsistent. Some sites report meaningful traffic from source links within Overviews. Others report negligible volumes. The pattern appears to correlate with how prominently the source is displayed and how much the user’s query requires more detail than the Overview provides. For complex topics where the Overview is necessarily incomplete, users click through. For simple queries where the Overview is sufficient, they do not.

The credibility signal is harder to measure and easier to overvalue. Being cited by Google does not translate into brand recall in the way that appearing at position one in traditional organic results does. Users reading an AI Overview are not in the same attentional state as users scanning a list of results and choosing which site to visit. The citation is peripheral, not central, to their experience.

Semrush’s analysis of Google AI Mode’s SEO impact is useful context here, particularly the sections on how source attribution works and what the click behaviour patterns look like. The picture is not catastrophic, but it is not the compensation story that some commentators have suggested either.

How to Measure the Actual Impact on Your Site

This is where most teams are currently failing, not because they lack the tools but because they have not set up the measurement correctly. Google Search Console now shows AI Overview impression data, but it requires deliberate segmentation to separate from standard organic data. If you are looking at aggregate organic traffic trends and attributing changes to AI Overviews without that segmentation, you are guessing.

The measurement approach I would use starts with identifying which of your ranking pages are appearing in queries that trigger AI Overviews. Search Console’s performance report allows filtering by query, and cross-referencing that with AI Overview prevalence data from tools like Semrush gives you a clearer picture of exposure. From there, you can track CTR trends on those specific pages over time, comparing periods before and after AI Overview rollout for those query types.

The second measurement priority is understanding what happens to users who do land on your site from AI Overview-adjacent queries. If your content is genuinely serving a need that the Overview could not fully address, you should see engagement metrics that reflect that. If users are arriving and immediately leaving, that is a signal that the content is not differentiated enough to justify the click in the first place.

I spent a significant part of my agency career trying to fix measurement frameworks that were producing confident-looking numbers built on shaky foundations. The problem was not a lack of data. It was a lack of honest interpretation. AI Overviews have introduced a new variable that requires recalibrating how you read organic traffic trends, and the teams that will handle this well are the ones willing to admit that their current dashboards do not yet tell the full story. Moz’s analysis of AI’s role in SEO and content marketing covers some of the measurement considerations worth working through.

The Strategic Response That Actually Makes Commercial Sense

There are two responses circulating in the SEO community that I think are wrong, and one that I think is right.

The first wrong response is panic-driven abandonment of informational content. The argument goes that if AI Overviews are going to answer informational queries, there is no point investing in content that targets them. This ignores the fact that AI Overviews are not universal, that coverage varies by query and by geography, and that informational content still serves functions beyond direct traffic generation, including establishing topical authority that supports ranking on commercial queries.

The second wrong response is ignoring the problem entirely and waiting to see what happens. I have watched businesses take this position on major industry shifts before, and it tends to end in a scramble to catch up rather than a considered adaptation. The traffic data is clear enough that doing nothing is a choice with consequences.

The right response is a portfolio rebalancing rather than a wholesale pivot. That means three things in practice. First, audit your informational content and identify which pages are most exposed based on query type and AI Overview prevalence. Prioritise those for either elevation (adding genuine depth, perspective, and specificity that AI cannot synthesise) or deprioritisation (reducing investment in content that serves intent AI can satisfy). Second, shift marginal investment toward content that serves intent stages closer to conversion. Comparison content, use-case specific content, content that requires product or service specificity, is harder for AI to replicate and more directly tied to commercial outcomes. Third, build content that is genuinely original in its perspective. This is harder than it sounds and requires writers who have something to say, not just the ability to produce well-structured text.

When I was growing an agency from around 20 people to over 100, one of the consistent challenges was convincing clients to invest in content that served a commercial purpose rather than a traffic purpose. The two are not the same thing, and AI Overviews are making that distinction more commercially consequential than it has ever been.

What This Means for Technical SEO and Structured Data

There is a practical dimension to AI Overviews that gets less attention than the content strategy conversation: how Google selects which sources to cite, and whether technical SEO signals influence that selection.

The evidence suggests that structured data implementation does influence AI Overview source selection, though not in a deterministic way. Pages with well-implemented schema markup, clear entity relationships, and strong E-E-A-T signals appear more frequently as cited sources. This is not a guarantee, and the selection algorithm is not transparent, but the correlation is consistent enough to be worth acting on.

The broader technical point is that AI Overviews have raised the floor on what “good enough” technical SEO looks like. A page that was ranking adequately with mediocre technical implementation may now be losing ground to pages that are better structured for machine interpretation. Canonical tag implementation, crawl efficiency, and entity clarity are not new concepts, but they matter more when Google’s systems are making rapid, automated judgements about which sources to surface.

The ongoing evolution of Google’s SERP architecture is worth tracking because the AI Overview rollout is not a static feature. The coverage, format, and source selection behaviour will continue to change, and teams that are monitoring those changes in real time will be better positioned than those treating it as a one-time adjustment.

The Longer View on Search Behaviour Change

It is worth stepping back from the immediate traffic impact conversation and considering what AI Overviews represent in terms of how users relate to search. The click-through rate suppression is a symptom of a more fundamental shift: for a growing share of queries, users are finding that they do not need to visit a website to get what they came for.

This is not entirely new. Featured snippets, knowledge panels, and local packs have been reducing clicks on certain query types for years. AI Overviews accelerate that trend and extend it to a wider range of queries, but the underlying dynamic is the same. Google has always been trying to answer questions rather than route traffic. The technology has just become good enough to do it more comprehensively.

The commercial implication for most businesses is that organic search traffic, as a metric, is becoming less reliable as a proxy for brand visibility or marketing effectiveness. A site that sees its informational traffic decline while its commercial query rankings hold steady may actually be in a better commercial position than its traffic numbers suggest. That is a measurement recalibration that most SEO reporting frameworks have not yet made.

I have always been sceptical of traffic as the primary measure of SEO success. When I was managing large paid search accounts, the discipline of connecting spend to revenue was built into the channel by design. SEO has historically lacked that discipline, and the result has been programmes that optimise for rankings and traffic without being honest about whether those metrics connect to business outcomes. AI Overviews are forcing that conversation in a way that is in the end healthy, even if it is uncomfortable in the short term.

If you are rethinking how your SEO programme is structured in response to these changes, the Complete SEO Strategy hub covers the strategic and tactical decisions that sit underneath the AI Overview question, including how to build a keyword architecture that holds up as the SERP continues to evolve.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions