AI Overviews Cut Your Paid Search CTR 68%. Here's What the Surviving Brands Did Instead.
Paid CTR on AI Overview queries has dropped 68%. The brands holding their ground aren't spending more, they're already being cited. Here's the three-week sequence to get there.
If you've been watching your paid search numbers lately and something feels off, you're not imagining it. Paid CTR on queries that trigger AI Overviews has dropped 68%, from 19.7% down to 6.34%, between June 2024 and September 2025, according to Seer Interactive data reported by Search Engine Land. The collapse wasn't gradual. In July 2025 alone, paid CTR fell from roughly 11% to 3% in a single month when Google expanded AI Overviews into commercial and navigational queries.
The instinct most marketing teams reach for: tighten targeting, raise bids, experiment with ad formats. That's a reasonable response to a performance problem. It's also the wrong one here, because the problem isn't in your campaign settings.
The brands holding their performance through this shift have something in common. Search Engine Land's analysis from February 26 puts it plainly: AI citation creates a winner-take-all dynamic. Brands consistently cited in AI Overviews don't just survive, they become disproportionately visible in the queries where everyone else is fighting over fewer clicks. Those brands aren't winning through better bidding. They're winning because AI systems already treat them as authoritative.
The question is how they got there, and what you can actually do about it this week.
The thing everyone gets wrong about AI Overviews
The common advice on getting into AI Overviews is technical: structured data, better heading hierarchy, FAQ schema, concise answers near the top of the page. All of that matters at the margins. None of it explains why some brands appear consistently and others don't.
Search Engine Land's February 12 analysis by Carolyn Shelby frames it accurately: AI search runs on entity mass, not surface signals. In AI systems, entities function like celestial objects, their influence is defined by mass, and mass is built through third-party citations, independent mentions, and corroboration across the web. On-page signals tell the system you exist. Entity mass tells the system you matter.
The difference matters. A brand can have technically perfect structured data and still not appear in AI Overviews if the AI system doesn't have enough external evidence to trust that brand as an authoritative source. Credibility has to be independently reinforced, not just asserted, before it gets used in a machine-generated answer.
This isn't a new concept dressed in new language. It's the same mechanism that made PR work before anyone had to think about AI: a placement in Forbes carries weight not because you asked Forbes to say you were credible, but because an independent editorial decision confirmed it. The AI system's calculation is similar. Muck Rack's analysis of over one million AI citations found 82% come from earned media sources, the AI is using what the broader web has already confirmed about you, not what you've said about yourself.
Where most brands are stuck right now
Here's what's actually happening in most marketing teams right now. They're running A/B tests on ad copy, experimenting with Performance Max, and trying to claw back the CTR they've lost. Some of this produces incremental gains. Meanwhile, the structural shift continues.
The issue is that paid CTR recovery is a short-term fix to a long-term problem. According to Adthena's data, AI Overviews now appear on 79% of longer finance queries and 84% of retail comparison and product discovery queries. That coverage is expanding, not contracting. Any category where buyers do research before purchasing is going to see this.
The brands that are pulling ahead aren't just weathering the CTR shift, they're using it to widen the gap. When your brand gets cited in the AI answer, two things happen. Your paid spend becomes more efficient because branded queries hold their performance. And you get organic visibility in the zero-click surface where your competitors aren't. You can read more about how this citation dynamic plays out for AI-era share of voice in AT's breakdown here.
The three-week sequence
This is the actual operational sequence. A full entity authority program takes months, but this is the work that moves the needle in the near term.
Week 1: Establish your entity baseline.
Before you can build entity mass, you need to know what the AI systems currently know about you. Run your brand name through ChatGPT, Perplexity, and Google's AI Overviews for the four or five queries where you want to appear. Screenshot what comes back. Note which brands are being cited, which publications are being used as sources, and whether your brand appears at all.
Then check your existing coverage. How many third-party publications, actual editorial outlets, not press release wire services, have published original coverage of your company, your product, or your team in the last 12 months? Count them. If the number is under five, that's your problem. If the publications are mostly tier-3 outlets or industry trade sites, that may also be your problem, AI systems weight source authority, not just source count.
Week 2: Build the citation layer that's missing.
The specific thing AI systems are looking for is third-party editorial coverage in publications they already index and trust. That means Forbes, TechCrunch, Business Insider, Bloomberg, industry-specific outlets with genuine editorial standards, publications that your buyers would also find credible.
This is where the execution splits from generic advice. The standard GEO recommendation is "get mentioned in authoritative publications." The actual work is building a direct editorial relationship with the journalists and editors who cover your category, and pitching in a way that earns coverage rather than just requesting it.
For teams doing this without dedicated PR support: your fastest path is executive thought leadership tied to a specific, timely angle. Find the one claim your CEO or VP of Marketing can make that is both true and counter-intuitive, something that pushes back against conventional wisdom in your category. Pitch it to two or three publications with a direct note to the relevant editor. Not a press release. Not a PR wire blast. A one-paragraph pitch from someone who can actually write and deliver the piece.
Coverage from one credible outlet starts the signal. Coverage from three or four in the same 60-day window starts building momentum in the entity model. AT's research on why earned media outperforms on-page optimization for AI citations explains the mechanism in more detail.
Week 3: Structure your owned content for machine extractability.
Once you have external citations starting to accumulate, your owned content becomes more useful, because now the AI system has a reason to look at it. The goal here is making the content easy for AI systems to read, extract, and use.
Practically that means: clear entity definitions on your About page and author pages (who you are, what you do, what category you operate in), structured content on your product pages that answers the specific questions buyers ask AI systems, and consistent entity references across every page, your company name, product names, and key executives referenced the same way every time.
The technical SEO playbook also applies: allow AI crawlers (GPTBot, ClaudeBot, PerplexityBot) in your robots.txt unless there's a specific reason not to. Blocking AI crawlers means the system can only form its opinion of you from third-party sources, with no first-party content to corroborate or expand on what it already knows.
What this is actually building
The three-week sequence gets you started. The thing you're actually building over the next six to twelve months is what you might call a citation infrastructure, a network of trusted editorial sources that have independently confirmed your brand's authority in the category you're trying to own.
This is the layer of PR strategy that's survived every algorithm change because it predates algorithms. Machine Relations is the discipline that formalizes it for the AI era: earned media in trusted publications, built through genuine editorial relationships, pointing back to your brand when AI systems do their first pass on buyer research. That mechanism, PR's original mechanism, is now what determines whether you appear in the AI answer or get written out of the search entirely.
The practical implication: spending more on paid search while your AI citation baseline is weak is spending into a losing position. The brands pulling ahead right now are solving both problems in parallel, but they started with the citation layer. The ads work better once the citation layer is there. The reverse doesn't happen on its own.
Run your brand through an AI visibility check before next week's budget conversation. The audit is free. It takes five minutes and gives you a clear read on where you actually stand before you decide where to put Q2 budget.