AI Overviews are dominating search, and now getting cited matters more than getting clicked. Let’s learn how to optimise for Google AI overviews and learn the best GEO practices that work in 2025.
Do you remember the first time you searched for something on Google and saw that AI-generated summary at the top? We do. We were looking for a way to write a creative blog post for a client. Instead of clicking through five different blogs, Google’s AI Overview gave us the exact strategy, considerations, and ideas about researching the topic.
That moment changed everything for us as content marketers. Your carefully crafted content is ranking well, but somehow, it’s not getting the clicks it used to. So, you need to change your strategy now!
Key Findings
The single most critical takeaway is that getting cited matters more than getting clicked.
- When a Google AI Overview appears, the top-ranking organic result sees its Click-Through Rate (CTR) plummet by 34.5% (from 7.3% to 2.6%). This means that even holding the number one organic spot is no longer a guarantee of traffic.
- The focus must shift to Generative Engine Optimisation (GEO), making your content visible and citable in the AI-generated answers.
- Users who do click through after seeing the AI summary are more qualified and engaged. By being the cited source, you establish authority and capture this higher-intent audience.
What Exactly Is GEO and Why Should You Care?
Did you know? AI-generated summaries now appear in 13% of all queries as of May 2025, which is double what it was just a few months ago. That’s the new normal for SEO and content marketers.
GEO, or Generative Engine Optimisation, is essentially the practice of making your content visible and citable in AI-generated answers. Think of it as SEO’s evolution for the age of ChatGPT, Claude, Perplexity, and Google AI Overviews.
But here’s where it gets interesting. Google’s Gary Illyes recently made a statement that confused a lot of marketers. He confirmed that specialised AEO or GEO optimisation isn’t necessary, and that standard SEO is all you need to rank in AI Search.
So does that mean you shouldn’t bother with GEO strategies? Not quite. What he’s really saying is that the fundamentals haven’t changed; you still need solid SEO. At Potens Digital’s SEO services, we apply those fundamentals but in a slightly different way to cater to the GEO and AEO needs.
Your Traffic Is About to Change
Let us be straight with you. When an AI Overview appears, organic click-through rates drop by 34.5%, with the top-ranking site seeing its CTR plummet from 7.3% to just 2.6%. That’s not a typo. Even if you’re ranking number one, you’re losing more than half your traffic when Google’s AI decides to summarise the answer.
But don’t panic because AI results may give people more context about a topic overall and display more relevant supporting links, which may provide a more engaged audience and new opportunities with visitors. The users who do click through are more qualified, more engaged, and more likely to convert.
According to a search engine expert at Potens Digital:
The focus is shifted from “get all the clicks” to “get the right citations and the right clicks.”
How AI Overviews Actually Work
AI Overviews read the entire internet but only remembers and cites the most trustworthy, well-structured sources.
The system works on Retrieval Augmented Generation, or RAG. Essentially, when someone asks a question, Google’s AI searches the web first, retrieves the top results, analyses them, and then generates a summary based on those trusted sources.
It is important to understand that if your content doesn’t rank organically in traditional search results, you won’t be retrieved for the AI answer. Traditional SEO is still the foundation. Everything else builds on top of it.
But here’s where things get interesting. 48% of citations in AI Overviews overlap with the top 100 organic search results, while the remaining 52% originate from sources that don’t appear in the top 100 ranked results. This means Google’s AI is casting a wider net than traditional search algorithms ever did.
How To Create Content That AI Models Love to Cite
If you want to get cited in AI overviews, you need to follow a path to make sure your content is being seen by AI, whether it is Google’s AI or LLMs.
1. Write Like You’re Answering a Friend’s Question
Our team learned this lesson the hard way. We used to write blog posts that were technically perfect, keyword-optimised, well-researched, and properly formatted. But they read like technical manuals. Then we started writing like we were explaining something to a friend over coffee.
This worked like a wonder. Not only did our engagement increase, but our content started appearing in AI Overviews more frequently.
Here’s how it works:
Question, Answer, Expand.
Start every major section with a clear question as your heading. Immediately follow with a direct, concise answer (think one or two sentences that could stand alone). Then expand with context, examples, and supporting details. Finally, reinforce the main point.
For example:
“How long should you brew cold brew coffee?
Cold brew coffee should steep for 12 to 24 hours in the refrigerator for optimal flavour extraction. The longer steeping time, compared to hot brewing methods, creates a smoother, less acidic taste profile that many coffee enthusiasts prefer.”
The AI can easily extract the direct answer, but curious readers get the additional context they’re seeking.
2. Structure Your Content Wisely
AI models are excellent at parsing well-structured content. They struggle with walls of text and buried information. So, use short paragraphs. Seriously, if your paragraph is longer than three or four sentences, break it up. This isn’t because readers have short attention spans, but because scannable content is more accessible to both humans and AI.
Bullet points are your best friend. When you have a list of benefits, steps, or features, format them as bullets or numbered lists. 82% of Google AI Overview citations come from deep pages rather than homepages, and only 0.5% of citations link to homepages. That means your detailed, well-structured blog posts and resource pages are more valuable than your homepage.
Tables work brilliantly for comparisons, specifications, and data. AI models can easily extract structured data from tables and present it in their summaries.
3. Add An FAQ Section
Old school, but the FAQ sections are experiencing a huge importance. They’re perfect for AI extraction because they present information in the exact format AI models prefer, “clear questions with concise answers.”
But don’t just create generic FAQs. Base them on actual questions your audience asks. Use tools like Answer the Public, Google’s “People Also Ask” section, and your own customer support tickets to find real questions people are asking.
A pro tip! Format each FAQ answer to stand alone. Don’t rely on context from previous questions. AI models often extract individual answers without the surrounding context, so each response needs to make sense independently.
E-E-A-T Is Still Very Important To Make AI Trust You
Google’s E-E-A-T principles (Experience, Expertise, Authoritativeness, and Trustworthiness) have become even more critical to compete in AI Overviews.
1. Show, Don’t Just Tell
Anyone can claim to be an expert. AI models are looking for evidence.
Include author bios that showcase relevant credentials and real-world experience. But go beyond just listing qualifications. Share specific examples of your work, case studies you’ve been involved in, or projects you’ve completed.
We started including a brief note in our articles about why we are qualified to write about the topic. For a piece on content marketing, we mentioned that we’ve managed content strategies for Fortune 500 companies and generated over $10 million in attributed revenue. That’s not bragging, as this is how you are establishing credibility.
Original data and statistics are incredibly valuable. If you can conduct your own research, surveys, or experiments, you’ve created something that AI models can’t find anywhere else. This makes you a unique, authoritative source.
2. Cite Your Sources
When you reference statistics, research, or claims, cite them. Link to reputable external sources. This does two things: it shows AI models that you’re building on trusted information, and it demonstrates that you’re not just making things up.
But be strategic about your sources. Link to academic papers, government databases, industry reports, and respected publications. Avoid citing questionable sources or competitors who might be gaming the system.
3. Transparency Builds Trust
Be honest about limitations, conflicts of interest, and what you don’t know. If you’re reviewing a product and received it for free, say so. If a topic requires specialised medical or legal advice, acknowledge that and recommend consulting professionals.
AI models are getting better at detecting authentic, trustworthy content. The days of purely promotional content dominating search results are ending.
Tweak Your Technical SEO Strategy For AI
Since we are living in the times of AI, we have to prepare the backend of our sites so that they are fetchable or scannable by AI. So, consider the following to be ready for AI crawlers.
1. Make Sure AI Can Find and Read Your Content
This seems basic, but you’d be surprised how many sites accidentally block AI crawlers. Check your robots.txt file to ensure you’re not blocking GPTBot (OpenAI’s crawler), Google’s crawlers, or Bingbot.
Speaking of Bing, here’s something many marketers overlook. Some AI models, including ChatGPT Search, rely on Bing’s index. If your content isn’t indexed in Bing, you’re invisible to those AI systems. Make sure you’ve submitted your sitemap to Bing Webmaster Tools.
Keep your critical content in clean HTML. Don’t bury important information behind JavaScript interactions or complex frameworks. AI crawlers need to access your content easily.
Your URLs matter more than you think. Use clean, descriptive URLs like /topic-location/ instead of /page?id=12345. AI models use URLs as context clues about your content’s relevance.
2. Schema Markup To Speak AI’s Language
Schema markup is structured data that tells search engines exactly what your content is about. Think of it as translating your content into a language AI models understand perfectly. Start with the Organisation Schema. This defines your brand entity, logo, and official social media profiles. It’s like creating a verified profile that AI models can reference.
For blog posts and articles, implement Article Schema. Specify the author, publication date, headline, and featured image. This helps AI models understand the context and freshness of your content.
FAQ Schema and HowTo Schema are incredibly powerful for getting featured in AI Overviews. They make your Q&A content explicitly clear to AI systems, removing any ambiguity about what question you’re answering. Schema is not technically difficult to implement, but most sites don’t use it properly.
Local SEO For AI Overviews
If you’re a local business, your Google Business Profile is potentially your most powerful tool for AI visibility.
Your GBP has roughly 90% influence on local AI responses. Make sure your address, phone number (including the correct area code), and business categories are accurate and complete.
Reviews are crucial. Not just having them, but the volume and velocity of reviews. AI models look at review patterns to assess reputation and relevance. If your competitors have 50 reviews and you have 5, you’re at a massive disadvantage.
Run a review gap analysis. Check what your competitors have and create a strategy to close that gap. Respond to every review, positive or negative, to signal active reputation management.
A strategic tip that many businesses miss! Don’t link your GBP to your homepage. Link it to the most relevant local landing page. If you’re a law firm with multiple locations, link your Charleston office’s GBP to your Charleston service page, not your general homepage.
In your GBP description, deliberately associate your brand with your core topics and location. For example: “Charleston’s premier personal injury attorney specialising in car accident cases and wrongful death claims.” You’re training AI models to connect your entity with specific topics and locations.
Building Authority Beyond Your Website
Your website is just one piece of the puzzle. AI models train on the entire internet, which means your off-site presence matters enormously.
1. Establish Your Digital Footprint
Think of this as teaching AI models who you are and what you’re about. The more consistent information AI systems find about your brand across the web, the more confident they become in citing you.
Get listed on Wikidata if possible. It sounds ambitious, but it’s actually achievable for many businesses and professionals. Wikidata influences knowledge panels and LLM training data.
The key is consistency. Your business name, address, phone number, and key details should be identical across all platforms. Inconsistencies confuse AI models and dilute your authority.
2. Quality Backlinks Still Matter
Earn backlinks from respected industry sites and publications. But in the AI era, it’s about association. When authoritative sites link to you, AI models learn that you’re part of the trusted network in your industry.
Pursue localised links through sponsorships, scholarships, local news coverage, and community involvement. These help AI models understand your local relevance and authority.
YouTube citations in Google AI Overviews have surged 25.21% since January 2025, particularly in the healthcare industry. This tells us that video content is becoming increasingly important for AI citations. Don’t ignore video as part of your content strategy.
3. Control Your Narrative with Strategic PR
Use press releases and media coverage proactively. When you win an award, open a new location, achieve a significant milestone, or have an interesting story, distribute a press release through reputable platforms.
The goal is to shape the narrative around your brand. You’re influencing what information is available for AI models to learn from. If all the information about you online is positive, current, and accurate, that’s what AI will present.
Measuring Success In The AI Era
Traditional metrics like clicks and impressions are becoming less useful. You need new ways to measure your AI visibility.
1. Track Citations, Not Just Rankings
Regularly audit which sources AI models cite for queries related to your industry. If you’re not appearing in those citations, you need to understand why. Are your competitors using better structured data? Do they have stronger E-E-A-T signals? Are they more frequently cited on authoritative third-party sites?
Create a list of target queries and test them across multiple AI platforms, Google AI Overviews, ChatGPT, Claude, and Perplexity. Track when your brand appears, in what context, and which sources the AI cites.
Monitor brand mentions even when they’re not linked. AI models often cite brands without providing direct links, or they summarise your content without attribution. Track these mentions to understand your actual AI visibility.
2. Use Synthetic Queries for Testing
Convert your target keywords into realistic, conversational queries. Instead of testing “best coffee maker,” try “I’m looking for a coffee maker that can brew a full pot in under 5 minutes for busy mornings. What are my best options?”
These longer, more natural queries better simulate how people actually use AI search tools. They’ll give you more accurate insights into your AI visibility.
3. Monitor Referral Traffic from AI Platforms
Set up tracking in Google Analytics to monitor referral traffic from chat.openai.com, perplexity.ai, and similar domains. Look for patterns over time, not just individual spikes.
Watch for increases in branded searches. Often, people discover you through an AI citation, then search for your brand directly. This indirect traffic is a key indicator of AI visibility success.
5 Common Mistakes To Avoid
Let me save you some pain by sharing mistakes we’ve seen (and made ourselves).
- Don’t keyword stuff. AI models are sophisticated enough to recognise when you’re gaming the system. Focus on semantic richness, using a variety of related terms and concepts naturally, instead of repeating the same keyword endlessly.
- Don’t ignore the basics. All the GEO strategies in the world won’t help if your site is slow, broken, or impossible to navigate. Technical SEO fundamentals still matter.
- Don’t forget about humans. You’re optimising for AI, but humans are your ultimate audience. If your content reads like it was written for robots, you’ve failed. The best GEO content is content that humans genuinely find valuable.
- Don’t neglect updates. Fresh content signals relevance and authority. Regularly update your existing content with new information, current examples, and updated statistics. AI models favour recently updated content.
- Don’t copy your competitors. Yes, analyse what’s working for them. But don’t just replicate their approach. AI models value unique perspectives and original information. Find your angle.
Conclusion
Around one in five Google searches in March 2025 produced an AI summary, and that percentage is climbing. You can either adapt or watch your traffic disappear. But adaptation doesn’t mean abandoning everything you know about SEO. It means evolving your approach to meet the new reality.
The good news? Most of your competitors are still figuring this out. If you start implementing these GEO practices now, you’re ahead of the curve.
Start with the low-hanging fruit; add schema markup, create FAQ sections, improve your GBP, and structure your content for easy extraction. Then work on the harder stuff: building authority, earning citations, and establishing your brand as an entity AI models trust.
