I’ve been building websites and watching Google’s algorithm changes for 17 years. The shift happening right now? It’s bigger than anything I’ve seen before.
The Numbers Don’t Lie
Here’s what’s actually happening with search. Google’s AI Overviews now appear in 18.76% of search results as of November 2024, up from basically zero two years ago. By March 2025, that number hit 13.14% of all queries, and it’s climbing fast.
But here’s the kicker: when AI Overviews appear, they reduce click-through rates by an average of 34.5%. NerdWallet is a perfect example, they generated 35% more revenue in 2024 but lost 20% of their website traffic.
The math is simple. More searches are getting answered without clicks. The traditional “rank #1 and get 30% of clicks” game is breaking.
I started noticing this with my own projects around mid-2024. Traffic patterns that had been stable for years suddenly weren’t. Pages that ranked #2 or #3 for competitive terms saw traffic drops, but conversions stayed steady or even improved.
What We Actually Need to Do
The thing is, everyone’s talking about “AI SEO” but most of it is just repackaged basic SEO with fancy names. Let me break down what actually works based on real data.
Stop Chasing Rankings, Start Chasing Citations
Research shows that 50% of links in ChatGPT responses point to business websites, even though platforms like Reddit and Quora perform best individually. The new metric isn’t “where do I rank?” It’s “am I getting cited by AI systems?”
I tested this with our AixonAI content. Instead of targeting “AI resume builder” (competitive, crowded), I started creating content around specific pain points like “why recruiters ignore technical resumes” and “how to explain career gaps in AI interviews.”
Result? These niche pieces get cited by ChatGPT and Perplexity regularly, even though they don’t rank #1 for anything.
Schema Markup Actually Matters Now
I used to roll my eyes at schema markup. Felt like busy work. Not anymore.
Studies show that LLMs grounded in knowledge graphs achieve 300% higher accuracy compared to those relying solely on unstructured data. Platforms like Perplexity, Claude, ChatGPT, and Gemini rely on schema markup to interpret and rank information.
Here’s what I did for our job search content:
{
"@type": "FAQPage",
"mainEntity": [{
"@type": "Question",
"name": "Why do technical resumes get rejected by ATS systems?",
"acceptedAnswer": {
"@type": "Answer",
"text": "ATS systems fail to parse technical resumes because they use complex formatting, technical jargon without context, and non-standard section headers."
}
}]
}{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [{
"@type": "Question",
"name": "Why do traditional SEO rankings matter less in 2025?",
"acceptedAnswer": {
"@type": "Answer",
"text": "AI Overviews now appear in 18.76% of search results and reduce click-through rates by 34.5% on average. The focus is shifting from rankings to AI citations and comprehensive content that answers user intent completely."
}
}, {
"@type": "Question",
"name": "How do I track AI citations for my content?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Build a system to test key topics weekly on ChatGPT, Claude, and Perplexity. Screenshot when your content gets referenced and note which specific passages get quoted. This data is more valuable than traditional ranking reports."
}
}, {
"@type": "Question",
"name": "What type of schema markup helps with AI search visibility?",
"acceptedAnswer": {
"@type": "Answer",
"text": "FAQ schema, HowTo schema, Organization schema, and Person schema are most effective. LLMs grounded in knowledge graphs achieve 300% higher accuracy, making structured data crucial for AI understanding."
}
}]
}Within three months, this FAQ content started appearing in AI Overview responses for resume-related queries.
Content Depth Beats Content Volume
The old playbook was publish daily, target long-tail keywords, hope something sticks. That’s dead.
88.1% of queries that trigger AI Overviews are informational. AI systems want comprehensive answers, not keyword-stuffed fluff.
I shifted our content strategy from 2-3 posts per week to one deep piece every two weeks. Each piece covers a topic completely, from beginner to advanced level, with examples, counterarguments, and real case studies.
The weird part? Our organic traffic stayed roughly the same, but engagement metrics went through the roof. Time on page doubled, bounce rate dropped by 40%.
The Specific Changes I Made
Here’s exactly what I changed in our SEO approach for one of our projects, and you can copy this:
1. Restructured Existing Content for AI Consumption
I went through our top 20 performing posts and restructured them using what I call “AI-ready formatting”:
- Each section can stand alone as a complete answer
- Added context to every claim with sources
- Included quantified data wherever possible
- Used FAQ schema for common questions
For example, our post “How to Write a Technical Resume” became:
# How to Write a Technical Resume That Actually Gets Interviews
## The Problem with Most Technical Resumes
According to a 2024 study of 500 technical recruiters...
## Section 1: The ATS Problem
Technical resumes fail ATS systems because... [complete explanation with data]
## Section 2: The Recruiter Problem
Even when they pass ATS, technical resumes get rejected because... [complete explanation with examples]Each section works as a standalone answer to specific questions recruiters and job seekers ask AI systems.
2. Started Tracking AI Citations Instead of Rankings
I built a simple system to monitor when our content gets cited by AI tools:
- Weekly tests of key topics on ChatGPT, Claude, and Perplexity
- Screenshots when our content gets referenced
- Notes on which specific passages get quoted
This data became more valuable than ranking reports. I can see exactly which content AI systems trust and reference.
3. Changed Our Link Building Strategy
Instead of pursuing “high authority” backlinks, I focused on getting mentioned in places AI systems learn from.
Quora is the most commonly cited website in Google AI Overviews, with Reddit coming in second place. So I started answering questions on these platforms with genuine, helpful responses.
Not “visit our blog for more info” answers. Real, complete answers that happen to reference our approach or data when relevant.
4. Added Structured Data Everywhere
I went beyond basic schema and added structured data for:
- FAQ sections using FAQPage schema
- How-to guides using HowTo schema
- Our company info using Organization schema
- Individual team bios using Person schema
This structured data helps search engines and AI systems understand the content of your website, and the results showed up within months.
Result: What Actually Happened
Six months after making these changes, here’s what changed:
Traffic: Down 15% overall, but up 40% for high-intent queries. The traffic we lost was low-quality anyway.
Conversions: Up 60%. People finding us through AI citations or deep content were much more likely to sign up.
AI Citations: Our content now gets referenced 3-4 times per week across different AI platforms. Six months ago, it was zero.
Engagement: Average session duration increased from 2:30 to 4:15. Bounce rate dropped from 68% to 41%.
The most interesting change? We started getting inbound leads from people who said they found us through “AI search” rather than Google. These leads converted at 3x the rate of traditional SEO traffic.
The Real Lesson Here
This isn’t about abandoning SEO. It’s about recognizing that search is expanding beyond Google’s ten blue links.
People are using ChatGPT, Meta AI, Claude, and other tools to find information. If your content only works for traditional search, you’re missing huge opportunities.
The companies winning right now are the ones creating content that works for both human readers and AI systems. They’re not trying to game algorithms, they’re trying to be genuinely helpful at scale.
Here’s what I wish someone had told me a year ago: stop thinking about SEO as ranking manipulation. Start thinking about it as information architecture. Make your content so clear, so well-structured, and so authoritative that both humans and AI systems can’t help but reference it.
The old SEO game was about gaming Google. The new game is about being the best source of information in your field. That’s harder to fake, but it’s also more sustainable.
And honestly? It’s more fun. Instead of chasing algorithm updates, you’re focused on helping real people solve real problems. The traffic and rankings follow naturally from that.
If you’re wondering how your website stacks up against these new AI search requirements, I can help. I’ll analyze your content structure, schema implementation, and AI citation potential to show you exactly where the opportunities are.
Contact me to get a full audit report of your website content and discover which changes will have the biggest impact on your visibility in both traditional search and AI-powered platforms.
