13 Issues That Modified, 4 Issues That Stayed The Identical

For years, Google has been shifting away from easy key phrase matching. However the inclusion of their Gemini giant language mannequin (LLM) took this to a different degree.

Pre-LLM methods may acknowledge that “Apple” the corporate was completely different from “apple” the fruit. They may join associated ideas. However they couldn’t really comprehend nuance, deal with advanced multi-step reasoning, or synthesize data throughout dozens of sources the way in which a human would.

LLMs modified that. Serps with an LLM layer perceive context at a depth that lets them interpret ambiguous queries, infer unspoken intent, and join disparate items of knowledge in methods earlier methods couldn’t.

So, once you search one thing advanced like “reward concepts from California underneath $50 for individuals who have already got all the things,” Google’s AI Mode (based mostly on the Gemini giant language mannequin) doesn’t simply search for pages with these actual phrases. As an alternative, it makes use of what’s referred to as “fan-out queries”, breaking out your query into a number of associated searches behind the scenes (on this case, 10 completely different searches).

These fan-out queries would possibly look like:

• “California artisan items underneath $50”
• “California meals merchandise budget-friendly”
• “Expertise items California reasonably priced”
• “Arduous to buy individuals reward concepts”
• “California-made sensible gadgets”
• “Presents for individuals who don’t need extra stuff”
• “California specialty merchandise funds”

Or, when somebody searches “finest laptop computer for video enhancing underneath $1000 that’s not too heavy”, trendy AI doesn’t simply search for key phrase matches. It genuinely understands the trade-offs between processing energy, portability, and funds constraints.

Google now understands the nuance of prompts and does the analysis legwork for customers, moderately than simply handing you an inventory of hyperlinks to click on by means of.

Searches that show clear buying intent or mention specific brands are less influenced by AI. Whether in Google’s AI Overviews, AI Mode, or within large language models (LLMs), when users already know what they want to buy or which brand they prefer, AI tools usually display traditional search results instead of summarizing or replacing them.

While related terms like AEO (Answer Engine Optimization) or LLMO (Large Language Model Optimization) are sometimes used, the idea is the same: ensuring your brand is visible, cited, and accurately represented within AI-generated answers.

For marketers, this represents a new “battleground” for visibility—one where AI assistants, not just search engines, influence consumer decisions. Some brands are surfaced and recommended as trusted authorities, while others are simply left out of the conversation.

That doesn’t mean GEO replaces SEO. In fact, it builds on the same foundation. The difference lies in where and how visibility happens. Search and AI results draw from similar content pools, but how they surface and summarize that content can vary dramatically.

To earn mentions and citations in AI-generated answers, marketers will need to adapt familiar tactics—not invent entirely new ones. Strategies like influencer marketing, reputation management, and structured, semantically clear content are seeing a resurgence.

[note]

A citation is when an AI attributes information to your content and includes a link to your site—usually for data, stats, or how-to content.

A mention is when your brand or product name appears in an AI-generated answer without a link, often in product recommendations.

Each trade—and each model—experiences AI visibility in a different way. However what AI understands about your model seemingly comes from the identical place it learns about ours: different individuals’s web sites.

At Ahrefs, most AI-generated solutions reference us with out citing our website straight. As an alternative, they draw from critiques, information articles, boards, some social media, and blogs, combining that with the AI’s built-in information to type their understanding.

Apparently, after we analyze potential attain throughout our Model Radar indexes (attain = mentions x quantity of key phrases), the solutions the place Ahrefs is straight talked about or cited carry out considerably higher in three of the six tracked AIs—particularly AI Mode, Gemini, and Perplexity.

The takeaway right here is that your broader on-line presence—and what others say about you—issues greater than ever for the way AI perceives and represents your model.

And show the topics you miss out on so you can create new content for them and/or contact cited sites to feature you, so you can boost your AI visibility.

Most manufacturers have already got some AI visibility merely inherited from their current natural presence in Google and the broader internet. We do too—almost 8,000 citations throughout Google’s AI experiences and LLMs—all achieved with zero AI-specific optimization, purely from conventional website positioning.

However I think most manufacturers haven’t began optimizing deliberately for AI visibility but. It’s nonetheless early days. Those that start now may have an actual benefit—time to check, be taught, and construct authority earlier than everybody else catches on.

The window gained’t keep open endlessly. As soon as AI visibility turns into a mainstream precedence, standing out will get a lot tougher.

As talked about earlier, optimizing for AI doesn’t actually imply adopting brand-new ways (assuming there even are any GEO-specific ones but). It’s extra about making use of current website positioning practices with AI search in thoughts and monitoring a unique set of metrics. Right here’s a fast video that will help you stand up to hurry on optimizing for AI-driven search.

Agencies and consultants rushed to offer GEO services, sometimes before fully understanding what it actually meant. It feels a lot like the early SEO days: plenty of hype, scattered experimentation, and very little consensus on what really works.

For example, for months, the main conversation on LinkedIn about LLMs and AI Overviews was that they were taking away website traffic. But more recently, I’ve started seeing the opposite perspective emerge, showing that AI referral traffic and organic search can actually grow together.

Just look at TikTok’s numbers—Gen Z is using it like a search engine, and many even prefer it over Google.

SEO has evolved from a “magic money tap” to a strong channel with new limitations. Sustainable growth now requires multi-channel resilience rather than Google dependency.

It’s a toothpaste-out-of-the-tube moment. The shift has happened, and there’s no putting it back. I think that in a few years, no one will care that AI was used to create content—it’ll be as normal as cars being built by robots.

Sure, maybe one day AI will replace some jobs. But right now, the real risk is not using it and falling behind those who do.

The key is to lean into your human strengths—insight, strategy, creativity, unique human experience—and let AI handle the repetitive parts. That’s how you make yourself harder to replace by AI.

One of the most valuable uses of AI in SEO today is content optimization. Tools like Ahrefs’ AI Content Helper can dramatically reduce the time it takes to identify search intent, find topical gaps, and compare your coverage against competitors.

Check out how my colleague Louise used that AI tool to boost traffic by 72%.

At Ahrefs, we see this in action every day. My colleague Louise regularly breaks news about AI Search—almost every week—and her work is talked about, cited, linked and shared throughout the industry. That’s what real, authentic expertise looks like.

• Finding good keywords for your blog that you haven’t targeted yet.
• Research international SEO expansion opportunities.
• Discover the top-ranking sites across any group of keywords.
• Get an idea of any unique, non-standard approaches to content from your competitors.

MCPs represent a shift from AI as assistant (helping you write after you do the research) to AI as analyst (doing the research, spotting the opportunities, and then helping you execute). For SEO, where data analysis is often the bottleneck between insight and action, this can change how SEO tools work forever.

Content material that matches what individuals are truly looking for performs finest.

AI methods, identical to conventional search engines like google and yahoo, are designed to serve customers by offering probably the most related solutions to their queries.

For instance, as proven beneath, conventional Google Search, AI Mode, and ChatGPT all interpret the phrase “trainers” in the identical manner.

What we used to name key phrases or search queries at the moment are sometimes called prompts on this planet of AI, however that doesn’t imply key phrases are not helpful.

Key phrases nonetheless play an essential function in displaying what individuals are interested by, serving to you measure curiosity, demand, and development developments for various matters you would possibly wish to goal.

And since we don’t but have dependable information on AI search volumes, key phrase search quantity continues to be a great way to estimate general demand throughout each conventional and AI-driven searches.

Key phrases additionally make it simpler to observe AI exercise by grouping and summarizing comparable prompts underneath a shared matter. Prompts individuals sort into AI instruments are sometimes lengthy and really particular to their scenario.

As an illustration, an extended immediate like “I’m attempting to develop my small enterprise weblog and want some dependable content material advertising instruments that may assist with key phrase analysis, social media scheduling, and monitoring outcomes—what are among the finest choices proper now?” covers a number of sub-topics (key phrases) inside a broader theme.

Each immediate nonetheless has an underlying matter—or set of matters. Even when AI instruments use question fan-out methods (breaking down advanced prompts into particular search queries), they’re finally organizing round core matters. Two prompts that appear utterly completely different can truly be about the identical factor at their core.

That’s why I just like the method that we took with Ahrefs Model Radar. It provides you two ranges of matter insights. You possibly can see the precise key phrase that triggered the AI question and optimize for that particular time period. You can too take a look at the broader “mum or dad matter” behind these key phrases to succeed in a wider viewers.

The way you arrange your content material, and the way simple it’s for readers to skim and discover key particulars, nonetheless play an enormous function in each visibility and readability.

Right here’s an instance: our website positioning pricing examine is among the most continuously cited pages by AI on our website, although it was revealed lengthy earlier than the entire AI craze started. I can say for positive that AI visibility wasn’t even a consideration after we created that web page. What guided its construction was simply optimizing for readability and fundamental on-page website positioning finest practices.

LLMs simply love construction and readability. It helps them minimize or chunk giant our bodies of textual content into digestible, small components which can be simple to reference.

• Branded web mentions. The instances of the brand being mentioned on any page on the web (linked and unlinked).
• Branded anchors. The anchor texts in the links pointing to a website.
• Branded search volume. Search volume of keywords that include a brand.

The takeaway is that building brand authority remains as important as ever, perhaps even more so in an AI-driven search landscape where trust signals matter tremendously.

The good news? You don’t have to start over. Strong brands, high-quality content, and solid SEO fundamentals still win. What’s changing is where that visibility appears and how you measure success.

Right now, there’s still an early-mover advantage. Most brands haven’t begun optimizing for AI visibility, so those who start experimenting, learning, and building authority today will be ahead when everyone else catches on.