Dave Davies: LLM SEO Shortcuts, Attribution Risks & Agentic AI
▼ Summary
– Agentic AI is transforming SEO strategies, requiring marketers to adapt to new generative engine optimization techniques.
– Shortcuts with LLMs (like ChatGPT) can harm content quality; better approaches involve training models on specific author styles for consistency.
– Technical SEO now includes ensuring Common Crawl (CCbot) can access prerendered pages, as it’s a key data source for LLM training.
– Attribution is a major challenge as AI Overviews reduce click-through rates, with solutions like tracking impressions in Search Console becoming critical.
– The future of SEO lies in prioritizing generative engines, leveraging tools like Bing for AI-powered placements, and preparing for agentic AI systems that automate decision-making.
The rapid evolution of agentic AI is transforming how search engines operate, creating both challenges and opportunities for SEO professionals. During a recent discussion with Dave Davies, Head of SEO at Weights & Biases and upcoming SMX Advanced keynote speaker, key insights emerged about adapting strategies for this new landscape.
One major concern is the growing reliance on LLM shortcuts for content creation. While AI tools offer efficiency, Davies warns against overuse without refinement. For example, simply dumping an article into ChatGPT for editing often strips away the author’s unique voice. Instead, he suggests training models on an author’s previous work to maintain consistency—a method that takes more time but yields higher-quality results.
Another critical shift is the move from traditional ranking signals to relevance in generative engines. Unlike classic SEO, where links played a dominant role, large language models (LLMs) rely on token prediction—determining the most probable next word based on training data. This means brands must ensure their presence across authoritative sources, not just their own websites, to influence AI-generated responses.
A surprising technical oversight Davies discovered involves Common Crawl, a dataset used by major AI models. Many prerendered pages were blocked from being crawled, limiting their visibility in AI training data. Fixing this requires updating bot access rules—a small but crucial adjustment for future-proofing SEO efforts.
Attribution is another looming challenge. With users increasingly relying on AI-generated answers without clicking through, tracking conversions becomes murky. Davies highlights that while Google is working on integrating AI impressions into Search Console, marketers must rethink how they measure success in an era where direct traffic may decline.
Looking ahead, Davies emphasizes the need to prioritize generative engine optimization (GEO) alongside traditional SEO. For instance, optimizing for Bing’s ecosystem could be advantageous, given its role in powering ChatGPT’s search capabilities. Similarly, structured content formats—like tables and cited sources—help AI models better understand and surface information.
His upcoming SMX Advanced keynote, “Agents of Change: SEO in the Era of Agentic AI,” will delve deeper into how autonomous AI systems will reshape marketing. These systems, capable of making utilitarian decisions (like booking travel or shopping), may require entirely new optimization strategies—potentially shifting focus from content to agent-based interactions.
As the industry navigates these changes, Davies’ advice is clear: adapt quickly, test relentlessly, and prepare for a future where AI doesn’t just answer queries—it makes decisions on behalf of users. The winners will be those who embrace this shift rather than resist it.
SMX Advanced takes place June 11-13 in Boston. Secure your spot to learn more about the future of SEO and AI.
(Source: Search Engine Land)
