Seo has gone through its changes. Most SEO advice you’ll find back in time was all about keyword density, backlink counts or a checklist of meta tags to tick off. Search for something like seo services and you’ll land on pages built entirely around that checklist, selling packages priced by keyword count. Meanwhile, the search behavior that click was supposed to capture has already moved somewhere else.
That’s not a knock on any one agency. It’s the entire model. And the data on how search actually works now makes the gap obvious.
The click is disappearing. The visibility isn’t.
Google’s AI Overviews are not a side feature anymore. They’re the default experience for a huge share of informational search.
- Position-one organic listings lose 58% of their expected clicks when an AI Overview appears on the page, according to Ahrefs’ December 2025 study of 300,000 keywords pulled from Search Console data.
- BrightEdge tracked AI Overviews appearing on 48% of monitored queries as of February 2026, up 58% year over year.
- SparkToro’s clickstream research puts the zero-click share of US Google searches at roughly 58%, meaning the majority of searches now resolve without a single visit to any website.
I don’t think this is a temporary dip that recovers once the “novelty” wears off, and I’d push back on anyone who tells you to wait it out. The mechanism is structural. Google is answering the question before the user reaches the results. That’s not a ranking problem you fix with more content. It’s a distribution problem.
Ranking #1 stopped meaning what it used to
Here’s the part most SEO reporting glosses over: visibility and traffic are no longer the same metric.
A page can be cited inside an AI Overview, shown to thousands of people, and generate almost no clicks. Pew Research’s 2025 tracking of Google searches found organic click-through rate drops from 15% to 8% once an AI summary appears. Being present in the answer counts for something. It just doesn’t count in your Google Analytics dashboard.
What actually predicts whether you get cited is different from what predicts whether you rank. Ahrefs’ 2026 analysis found external brand mentions correlate with AI Overview citation at 0.664, one of the strongest signals in the data. Domain strength matters too: sites with more than 32,000 referring domains were 3.5 times more likely to get cited by ChatGPT than smaller sites.
My honest take: this rewards exactly the kind of authority-building that good SEO always claimed to reward, and quietly punishes the sites that were gaming rankings without ever earning real citations elsewhere. The AI systems didn’t invent a new game. They just stopped being fooled by the old one.
The businesses adapting are running two systems, not one
Splitting “SEO” and “AI search optimization” into separate line items doesn’t make sense anymore, and I’d argue against anyone who tries to sell them to you separately.
Semrush’s mid-2025 study of high-value B2B and marketing topics found visitors arriving through AI citations convert at 4.4 times the rate of standard organic visitors. That’s the number that should change how you prioritize. AI referral traffic is still small in absolute terms — Conductor measured it averaging just 1.08% of sessions across nearly 14,000 domains — but the visitors who do arrive that way have already done their research and are closer to a decision.
That changes what “winning” looks like:
- Fewer, deeper articles that are worth citing beat a high volume of thin posts.
- Structured, comparison-style content earns more citations. AirOps found comparison pages with data tables earned 25.7% more AI citations than plain prose pages.
- Technical accessibility for AI crawlers matters as much as it ever mattered for Googlebot. If an AI system can’t parse your page cleanly, it can’t cite you, no matter how good the content is.
Where automation actually helps, and where it doesn’t
AI tools are genuinely useful for the mechanical parts of this shift: auditing which pages already get cited, tracking brand mentions across the web, restructuring content for extractability, monitoring AI Overview appearances alongside traditional rankings. Automating that reporting layer is the highest-leverage move most teams haven’t made yet, because it’s tedious and nobody wants to build it by hand.
What automation doesn’t fix is the underlying content quality problem. Google’s own guidance on AI-generated content is explicit that scaled, low-value pages violate its spam policies regardless of how they were produced. Mass-producing generic explainers with AI, hoping volume compensates for depth, is the fastest way to end up with neither rankings nor citations.
Stop optimizing for the old scoreboard
If your reporting still leads with rankings and raw traffic, you’re measuring a game that’s already changing shape underneath you. Track AI citations next to your keyword positions. Build the brand mentions that feed both. And if an agency pitches you a package priced purely on keyword count, ask them what they’re doing about the 58% of clicks that AI Overviews are already taking away. If they don’t have an answer, you already know what you’re paying for.
