SEO isn't dead. It's not even close to dead. Here's what actually happened: Google added an AI layer on top of its existing search results. ChatGPT, Perplexity, and Gemini all run queries against Google (a process called query fan-out), pull the top-ranking pages, and synthesize answers from them. The input to every AI answer is a Google search result. Your Google rankings are your AI visibility. What changed is the format, not the fundamentals. Zero-click searches hit 58.5% in the US (SparkToro/Datos, 2024). AI Overviews doubled from 6.49% to 13.14% of results in two months (Semrush/seoClarity, 2025). Organic click rates fell from 44.2% to 40.3% year over year. Those numbers are real. But they don't mean what the "SEO is dead" crowd wants them to mean.
If you read those numbers and think everything's fine, you're wrong. If you read them and think SEO doesn't work anymore, you're also wrong. The truth is more specific and more useful than either take.
Certain SEO tactics are losing ground. Certain query types are worth less than they were. The way we measure success has to evolve. But Google still holds 89.62% of global search (StatCounter, 2025), search still sends more traffic than any other channel, and every AI search platform depends on Google results as its raw material. The businesses that understand this mechanism will capture a disproportionate share of what remains.
Here are five shifts grounded in what the data actually shows.
Shift 1: Target queries AI Overviews handle poorly
AI Overviews trigger on 88.1% of informational queries (Semrush/seoClarity, 2025). They're decent at simple factual questions. They're bad at everything else.
Queries where AI Overviews consistently fall short:
Subjective evaluations. "Best project management tool for remote teams of 5-10 people" requires judgment that depends on context AI can't capture from a snippet. These queries still need a human who's actually used the tools.
Multi-step processes. "How to set up Google Analytics 4 with a custom e-commerce data layer" needs sequential, detailed instructions that don't compress into a paragraph. Users click through because the AI answer isn't enough.
Queries with legitimate disagreement. "Should startups use microservices or monoliths?" has no single correct answer. AI Overviews either pick a side and lose credibility or hedge and become useless.
On March 31, 2026, a clean-room reimplementation of Claude Code hit GitHub and collected 50,000 stars in two hours. Alongside the Rust port, the actual npm package (@anthropic-ai/claude-code) has been sitting in node_modules since release, minified but readabl
AI referral traffic grew 527% between January and May 2025. ChatGPT referrals convert at 15.9%, nine times Google organic's 1.76%. And 76.1% of AI Overview citations come from pages already ranking in Google's top 10.
Those three numbers tell you everything yo
More than half of all Google searches in the US now end without anyone clicking a single result. According to SparkToro and Datos, 58.5% of US Google searches in 2024 were zero-click. For every 1,000 searches, only 360 clicks went to the open web.
That number
Recency-dependent queries. "Latest Google algorithm update" changes weekly. AI Overviews regularly cite outdated sources for fast-moving topics.
Experience-based queries. "What's it like to switch from Shopify to WooCommerce?" requires someone who actually did it. AI can't fabricate that convincingly.
Build your content calendar around these gaps. Use the SEO analyzer to identify which of your target queries currently trigger AI Overviews and which don't. The ones that don't are your immediate opportunities.
For a detailed breakdown of AI Overview query patterns and CTR impact, see our AI Overviews data analysis.
Shift 2: Build brand authority so AI cites you
When your brand gets cited in a Google AI Overview, you earn 35% more organic clicks than uncited brands (Semrush/seoClarity, 2025). Being cited is becoming the new ranking number one.
But here's the mechanism that matters: AI doesn't independently evaluate your brand. It cites what already ranks well in Google. When ChatGPT answers a question, it runs a Google search behind the scenes, pulls the top results, and synthesizes from those. If you're not ranking, you're not getting cited. Brand authority for AI citation starts with brand authority in Google.
What builds the kind of authority that gets you cited:
Original research. Publish data nobody else has. Surveys, analyses, proprietary metrics. Traditional rank trackers miss around 22% of total search visibility because they ignore AI citations, featured snippet ownership, and Knowledge Panel presence. That kind of original finding is exactly what AI systems pick up and cite.
Consistent topical focus. If you write about SEO, e-commerce, and cooking recipes, you're not an authority in any of them. Topical authority means depth, not breadth.
Named experts. Content with clear human authorship from people with verifiable expertise gets cited more often. Anonymous posts from "Admin" don't.
Entity presence. Google needs to understand your brand as an entity in its Knowledge Graph. That means consistent structured data across your site, an active Google Business Profile, citations from authoritative sources, and Wikipedia presence if you're notable enough.
Brand building takes time. Start now. The gap between cited and uncited brands will widen as AI Overviews expand, and the mechanism (Google rankings feeding AI answers) means this compounds over time.
Shift 3: Invest in E-E-A-T signals that actually work
Google's E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) has been in their quality guidelines for years. It matters more now because these signals determine which sources get pulled into AI-generated answers. Since AI platforms use query fan-out against Google, E-E-A-T doesn't just affect your rankings. It affects whether you show up in AI answers at all.
Most sites treat E-E-A-T like a checkbox: add an author bio, stick "reviewed by an expert" on the page, done. That's not enough.
E-E-A-T that moves the needle:
Experience. Show first-hand experience with the topic. Screenshots of you using the tool you're reviewing. Case studies from actual clients. Before/after data from real campaigns. There's a detectable difference between "I researched this topic" and "I did this thing," and both Google's quality raters and AI synthesis engines favor the latter.
Expertise. Author pages with real credentials linked to external verification (LinkedIn, professional associations, published work). Not just "John is a marketing expert with 10 years of experience" but specific, verifiable claims.
Authoritativeness. External signals: who links to you, who cites you, who mentions you. You can't shortcut this. You build it through original research, industry participation, and consistent quality over time.
Check how your pages score on these signals with the AI readiness tool. Pages that score well on E-E-A-T indicators are more likely to maintain traffic as AI features expand.
Shift 4: Optimize for AI search platforms directly
Google isn't the only AI answering questions. ChatGPT, Gemini, Claude, and Perplexity all pull from web content to generate answers. Each has its own crawling and citation behavior, but they all share one thing: they start with Google results.
This is the query fan-out mechanism in practice. When someone asks ChatGPT a question, it doesn't search its own index. It searches Google, pulls the top results, and synthesizes an answer from them. Perplexity does the same. So does Gemini. Your Google rankings are the foundation of your AI visibility across every platform.
What to do about it:
Publish an llms.txt file. This emerging standard (similar to robots.txt) helps AI crawlers understand your site structure and which content is available for citation. It's simple to set up and signals that you're participating in the AI search ecosystem.
Don't block AI crawlers. Some publishers are blocking GPTBot, Google-Extended, and other AI crawlers. Unless you have a strong reason and a business model that doesn't depend on discoverability, this is counterproductive. Being invisible to AI search means losing an increasing share of search interactions.
Structure content for extraction. Use descriptive headings, put key facts early in paragraphs, use schema markup to identify entities and relationships. LLMs strip HTML to plain text before processing, so structure primarily helps Google (which feeds AI answers via query fan-out) rather than AI systems directly. But well-structured pages rank better in Google, which means they get cited more in AI responses. Validate your markup with a schema validator.
Monitor your AI visibility. Track whether your brand appears in AI-generated answers across platforms. Google Search Console shows some AI Overview data, but for ChatGPT, Gemini, and Perplexity visibility, automated tooling is still immature. Manual prompt testing (searching your key queries in each platform) remains more reliable than any automated tracker. For a realistic look at where AI visibility fits relative to traditional SEO, see our GEO vs SEO analysis. For what AI can and can't handle in SEO workflows today, see our breakdown of agentic SEO capabilities.
Shift 5: Measure visibility beyond traditional rankings
Here's where most SEO teams are behind. They're still measuring success through keyword rankings and organic sessions. Those metrics still matter, but they're increasingly incomplete.
The zero-click data makes this clear. If 58.5% of searches end without a click, and AI Overviews cut organic CTR by 61% when present (Semrush/seoClarity, 2025), then a page can hold its number one ranking while losing most of its traffic. Rank alone doesn't tell you what it used to.
Metrics that matter now:
Impressions, not just clicks. Google Search Console impression data shows how often your pages appear in results, regardless of clicks. A page with 100,000 impressions and 2,000 clicks is still building brand awareness, even if the click rate looks low.
AI citation rate. How frequently does your content get cited in AI Overviews, ChatGPT responses, and other AI platforms? This is becoming the new "ranking." Automated tools for measuring it are still early-stage, so expect manual spot-checking to stay part of your workflow.
Featured snippet ownership. For your target queries, do you own the featured snippet? This data lives in Search Console under the search appearance filter. Snippets feed AI Overviews directly.
Brand search volume. Are more people searching for your brand over time? Brand search is the most resilient traffic source because Google rarely intercepts navigational queries with AI Overviews.
Share of voice across AI platforms. Beyond Google, is your brand showing up in ChatGPT, Gemini, and Perplexity when users ask about your category?
Total search visibility. A composite metric combining rankings, featured snippet ownership, AI citations, and Knowledge Panel presence. Traditional rank trackers miss about 22% of total search visibility by ignoring these factors. The AI citation component is the hardest to measure accurately right now, but tracking snippets, Knowledge Panels, and brand SERP features is straightforward. Tools like Ooty Analytics help capture what traditional rank trackers leave out.
What hasn't changed
Five shifts is a lot to take in. But before you overhaul everything, here's what still works exactly as it always has.
Technical SEO fundamentals. Fast sites, clean crawl paths, proper indexation, mobile usability. These are still table stakes. Google needs to crawl and understand your site before any AI system can cite it.
Quality content. The bar is higher, but the principle hasn't moved. Content that genuinely answers the user's question better than alternatives still wins. AI systems cite it precisely because Google ranks it.
Link building. Backlinks remain a core authority signal. AI systems like ChatGPT and Perplexity cite what ranks well in Google (they use query fan-out to search Google and synthesize results). So backlinks help you get cited in AI answers, but the causation runs through Google rankings, not through a separate AI authority signal. Build links to rank in Google, and your AI visibility follows.
Long-tail keywords. Specific, lower-volume queries are less likely to trigger AI Overviews and still convert at high rates. The long tail isn't dead. It's more valuable relative to head terms than it used to be.
User experience. Once someone lands on your page, the experience determines whether they stay, convert, and return. AI Overviews don't change that.
The bottom line
Google holds 89.62% of global search (StatCounter, 2025) and 93.82% on mobile. Search isn't going away. AI search depends on Google search. That's not speculation. That's how the systems work: query fan-out feeds Google results into every major AI platform.
The 14% traffic decline Semrush measured for non-news publishers isn't evenly distributed. Some sites are growing through all of this. The difference is strategy: the growing sites target the right queries, build genuine authority, produce content AI systems want to cite, and measure success in ways that reflect how search actually works today.
The five shifts above aren't theoretical. They're grounded in the data and the mechanism. Start with whichever one addresses your biggest gap, measure the impact, and iterate.
