Google’s Mueller: Vibe Coding Won’t Replace SEO Strategy

▼ Summary
– AI coding tools can quickly build functional websites, but achieving good SEO requires specific technical direction, similar to instructing a human developer.
– Vague instructions to AI, like “add some SEO,” produce vague results; Mueller got better outcomes by specifying domain, canonical setup, sitemap, and robots.txt from the start.
– Mueller builds test sites using Hugo and Firebase, recently switching to command-line AI tools like Claude Code and Gemini CLI.
– Both Mueller and Splitt note that technical understanding is crucial, as AI will make assumptions about site structure without clear guidance from the user.
– Vibe-coded sites can produce reasonable HTML that is not easily recognizable, but risks include AI-generated content and issues like crawlability problems and content hidden in JavaScript files.
Google’s John Mueller and Martin Splitt recently explored the intersection of AI-assisted coding and search engine optimization on the Search Off The Record podcast. Their takeaway? While vibe coding can quickly generate functional websites, it does not automatically deliver SEO success. Without clear, technical direction, the results are as hit-or-miss as handing a project to a developer who knows nothing about search.
Mueller compared the experience of asking an AI to “add some SEO” to asking a non-specialist developer the same thing. Vague instructions, he noted, lead to vague outcomes. “You can always tell the AI system, now add some SEO to it,” Mueller said. “But how that works out is if you go to a developer and add some SEO and it’s like, what do you mean. Sprinkle some meta tags and add some structured data.” He found that providing upfront specifics,such as the domain name, canonical URL structure, sitemap files, and a robots.txt,yielded far better results. He also verified that pages used sensible HTML, linked correctly, and set up pre-publish checks to confirm URLs returned content and JavaScript files weren’t blocked.
Mueller has been building test sites to study how Googlebot handles requests, deploying them on Firebase hosting with Hugo as a static site generator and GitHub for version control. He recently shifted from VS Code with Copilot to command-line tools, naming Claude Code and Gemini CLI as his current choices. Splitt, meanwhile, used Google AI Studio to build a client-side JavaScript tool. While the output was clean and readable,resembling a standard Next.js app,he ran into a frustrating loop where the AI stubbornly used a library he didn’t want. “I asked it for half an hour. I tried to make it not do what it wanted to do, and want to do what I wanted to do,” Splitt said. “And that was weird.”
Both acknowledged the tension at the heart of vibe coding’s promise: that you don’t need to know how to code. Mueller emphasized that technical understanding remains critical at every stage. Knowing what kind of site generator you want and how to structure pre-publish checks produces better results. Without that background, the AI makes assumptions. It might pick a static site generator, a JavaScript-heavy setup, or a full CMS with a database backend. “All of these are reasonable assumptions where if you talk to a developer they will also make these assumptions,” Mueller said. “But if you just tell the AI system like I want a website, then it will pick one.” For personal projects and low-risk static sites, the stakes are low enough to experiment. But for anything involving user data or a production service, Mueller added that you’d want someone who understands what they’re doing.
Regarding search visibility, the sites Mueller built produced reasonable HTML that wouldn’t stand out as vibe-coded. “Practically speaking, nobody can really recognize as being like, this is a vibe coded website,” he said, though he noted that common vibe coding frameworks can leave recognizable patterns. He also flagged a related risk with content. Once a site looks polished, it’s tempting to have the AI write the content too. Mueller acknowledged the tool can do that but said it’s not where he sees the most value. Splitt agreed, pointing out that AI-written content raises the question of why someone would visit a site instead of talking to the AI directly.
Mueller has previously identified gaps in vibe-coded sites. After reviewing a vibe-coded Bento Grid Generator on Reddit, he spotted issues with crawlability, obsolete meta tags, and content stored in JavaScript files that search engines couldn’t access.
The podcast did not offer formal guidance or policy positions on vibe-coded sites. Mueller and Splitt were simply sharing what they’ve tried and what they’ve run into. For people testing these tools, the message is clear: AI can handle parts of code generation well, especially for lower-risk projects. But it does not make SEO decisions on its own. Those still require someone who knows what to ask for.
(Source: Search Engine Journal)



