Google's Danny Sullivan stated on the "Search Off the Record" podcast that Google advises against segmenting content into "bite-sized chunks" specifically to rank well in Large Language Models (LLMs).
Key Statements
Sullivan indicated that Google does not want content creators to craft material solely for specific ranking systems, including LLMs. He mentioned discussions with Google engineers regarding this stance.
He suggested that while this approach might offer a minor advantage in some current LLM systems, it is not sustainable in the long term. Sullivan warned that efforts invested in optimizing for temporary system behaviors might become obsolete as ranking systems evolve to favor content written for human users.
Core Recommendation
The central recommendation emphasized writing content primarily for users. Sullivan believes that as ranking systems improve, they will increasingly reward content that serves user needs, aligning with what LLMs will ultimately value.
This guidance echoes long-standing advice within the Search Engine Optimization (SEO) community to prioritize user experience over attempts to manipulate algorithms.