“SEO is dead. GEO replaced it.”
What we know: Google said normal SEO practices are what get content into AI Overviews. GEO is a layer on top of SEO, not a replacement.
Source: Gary Illyes (Google), July 2025We checked each claim against a dated source. When new data changes the verdict, we update it here.
What we know: Google said normal SEO practices are what get content into AI Overviews. GEO is a layer on top of SEO, not a replacement.
Source: Gary Illyes (Google), July 2025What we know: Google said it does not use llms.txt. In Ahrefs data covering 137,000 sites, 97% of published llms.txt files got zero requests in May 2026.
Source: Google (2025), Ahrefs (2026)What we know: In Lily Ray's study of 100 B2B queries, Google cited a brand's own list without recommending that brand in 224 of 323 cases (69%). Sites scaling these lists lost 30% to 50% of visibility in January 2026.
Source: Lily Ray, Feb. and June 2026What we know: Engines split one question into many sub-searches (query fan-out). You need topic clusters, and a page per sub-search risks looking like spam.
Source: Mike King (iPullRank)What we know: AI answers change by person and prompt, so there is no fixed position. Measure visibility share and citations across a fixed prompt set.
Source: Stradiji measurement methodWhat we know: Models look up sources again on every question. In our five-week tracking, a wrong founding year showed up in four of five engines every week. Fix the third-party sources that carry the error.
Source: Stradiji tracking, 2026What we know: The Princeton GEO study measured visibility gains of up to 40% from these methods, but in a test environment. On live engines, the effect varies by industry and query.
Source: Aggarwal et al., KDD 2024What do AI engines remember, and what do they look up live? You will see why most of good GEO is good SEO.
What you'll learn: What GEO is, how it relates to SEO and AEO, and the three-door model this roadmap follows.
What you'll learn: Crawling, indexing, topical authority, and why more and more searches end without a click.
What you'll learn: What a model remembers from training versus what it looks up live, and how one question fans out into many searches.
An engine has to reach you, then recognize who you are, then believe what it finds. If one door is shut, the others don't help. Each door has a matching section on the GEO Score Card.
What you'll learn: AI crawlers, robots.txt, CDN and Cloudflare settings, and pages that only render with JavaScript.
Check this door on the score cardWhat you'll learn: Entities, Wikidata, Organization schema and consistent facts, so the engine stops guessing about you.
What you'll learn: Independent mentions, reviews, digital PR, and why sources that agree beat anything you say about yourself.
The GEO Score Card checks the same doors in about 10 minutes. Then come back and pick up where you left off.
Once the doors are open, the work is content worth quoting and measuring change the right way. No rankings, only visibility share.
What you'll learn: First-hand data, answer-first paragraphs, clear question headings and content no one else could have written.
What you'll learn: How often you are mentioned and cited across the same set of questions, every month, and how to separate traffic from revenue.
Agents that buy, ads inside AI answers, and the tools that make this work sustainable.
AI agents that browse and buy on a person’s behalf, and why accurate product data matters.
What ads in ChatGPT and Google AI Mode mean for the mentions you earn.
Tools to measure and fix AI visibility, with a free option next to every paid one.
New findings land in the newsletter first, and the roadmap is updated from it.
Every new finding tests the last one. GEO is a layer on top of SEO, and what we know about that layer changes fast. Where we were wrong, we pulled our own advice. Every claim on this page carries a date and a source.
Studies we read, tested and brought into the roadmap, with what each one found and which stage it touches.
We run our SEO and GEO projects with Claude working like a teammate. Each day has one focus, every fact lives in a single source of truth, and nothing changes on a live site without a human sign-off. Semrush, Search Console, GA4 and Clarity data flow straight into Claude through MCP connections. We show the setup step by step on YouTube.