The question on everyone's mind in marketing departments today isn't just "How do we rank on Google?", but "Why isn't ChatGPT mentioning us?" There's a common belief that if you do good technical SEO for Google, you'll automatically be favored by Large Language Models (LLMs). However, the reality is more complex. A recent study by Semrush, after analyzing 5 million URLs cited by ChatGPT Search and Google's AI Mode, sheds light on this fragmentation of the traditional search engine. Spoiler alert: technical SEO isn't dead, it's mutated into the foundation of what we now call AI Visibility. The end of the "ten blue links" The search landscape is becoming increasingly fragmented. Although Google maintains overwhelming dominance (90.06%), its slight but steady decline coincides with the meteoric rise of AI platforms. In this new "discovery" ecosystem, the playing field is clear: ChatGPT leads with 80% of the AI ​​search market share, followed by Perplexity (11%) and a rapidly growing Microsoft Copilot (5%). For SEO teams, this means that traffic forecasts based solely on Google are falling short. One of the most interesting concepts introduced by the study is that of the "informed clicker." Unlike the user who jumps from link to link on Google, the user arriving from an AI is already "informed." After all, the user who has already been assisted by AI knows who it is and why it's relevant. This translates into longer sessions, lower bounce rates, and a higher conversion rate. Semrush data confirms that pages cited by AI have better engagement metrics (visit duration and conversion). This doesn't mean that AI uses engagement as a direct ranking factor, but rather that pages that satisfy users tend to have the technical structure and authority that LLMs look for. What do the pages that AI chooses as sources have in common? The study highlights three key elements: 1. Structured data (schema markup) is the language of AI. Although AI can read almost anything, Schema Markup acts as a high-fidelity translator. The cited pages show a high implementation of: - Organizational structured data (up to 34%) - Article structured data (up to 26%) - Navigation path structured data (up to 20%) Interestingly, Google's AI Mode is much more demanding with this data than ChatGPT, more frequently citing pages that have implemented FAQ and LocalBusiness schemas. 2. Slug size does matter. Not too short, nor a hieroglyphic of parameters. The study reveals that URLs with slugs between 17 and 40 characters (especially in the 21-25 character range) receive the most citations. Extremely short URLs (homepages) or excessively long URLs packed with keywords are less successful in this new environment. 3. The Rise of AI Bots in Your Logs Traffic from OAI-SearchBot (OpenAI's crawler) is skyrocketing. This forces SEOs to take on a new task: managing their AI crawl budget. Monitoring log files to see which sections of your website are of most interest to these bots is now as critical as it was (and still is) for Googlebot. AI Visibility Has Become a KPI Measuring clicks and impressions is no longer enough. Modern strategy demands measuring AI Visibility: - Brand Mentions: Do we appear when someone asks for a recommendation in our category? -Citation frequency: How often are we cited as the source of truth by the model? Citation quality: Does it position us as leaders or as a secondary option? Let's be honest, technical SEO doesn't guarantee appearance in the AI, but you do have to do it well to have a chance of appearing. Without a clear structure, optimal page load speed, and structured data that defines who we are, we'll be invisible to the algorithms that are redefining the internet.