I Asked a 15-Year SEO Veteran How to Rank in ChatGPT | Dan Petrovic (DEJAN AI)

<p>Want your brand to show up in ChatGPT, Claude, and Gemini?</p><p>Here&#39;s the uncomfortable truth: LLMs don&#39;t crawl the web when they answer your questions. They search traditional indexes — ChatGPT leans on Bing, Gemini uses Google, and Claude appears to pull from Brave. So if you&#39;re not ranking in those indexes, you&#39;re not even in the consideration pool.</p><p>Dan Petrovic is one of the most respected (and delightfully nerdy) figures in SEO. He runs the Australian agency Dejan and was the top-rated speaker at Ahrefs Evolve in Singapore, where he presented research from asking Gemini one million questions to map how it ranks brands.</p><p>In this episode, Tim and Dan unpack what it actually takes to show up in AI search — from the mechanics of how LLMs select sources to the practical steps you can take today.</p><p>You&#39;ll learn the difference between grounding, citations, and mentions (and why a linked mention is the holy grail), why Reddit gets rejected over 90% of the time, why there&#39;s no such thing as &quot;prompt volume,&quot; and how Dan uses on-page optimization in an agentic loop to make models &quot;unable to resist&quot; a client&#39;s page.</p><p>In this episode:</p><p>(00:00) Intro<br>(00:48) My brand doesn&#39;t show up in ChatGPT — what do I do?<br>(04:02) Two ways to get into AI answers: outreach vs. building your own pages<br>(07:11) Is LLMs.txt actually pointless?<br>(09:29) How ChatGPT, Gemini, and Claude each use different search indexes<br>(10:42) Universal ranking factors: topical centrality, engagement, and popularity<br>(17:17) Practical example: what Ahrefs Podcast should actually do<br>(19:31) Should you optimize for every search engine, or just Google?<br>(22:59) How is AEO actually different from SEO?<br>(28:03) Can you influence LLM training data?<br>(33:58) Going nerdy: separating quality content from spam with Gzip compression<br>(37:27) On-page optimization in an agentic loop: making models &quot;unable to resist&quot;<br>(42:45) How often do LLMs ground their answers vs. answer from memory?<br>(47:37) Measuring AI visibility: entities, mention share, and citation share<br>(51:44) Why &quot;prompt search volume&quot; doesn&#39;t exist — and what to track instead<br>(56:35) How to generate prompts for tracking AI visibility</p><p>Where to find Dan Petrovic:<br>LinkedIn: linkedin.com/in/dejanseo<br>X: @dejanseo<br>Website: dejan.ai</p><p>Where to find Tim:<br>LinkedIn: linkedin.com/in/timsoulo<br>X: @timsoulo<br>Website: timsoulo.com</p><p>Referenced:<br>Dejan — dejan.ai<br>Brave Search — search.brave.com<br>Ahrefs — ahrefs.com</p>