The New Rules of AI Search

5One Strategy Doesn't Fit All—Content Strategies for Every Type of LLM Response

Here's something a lot of marketers get wrong about AI search: they treat every LLM like it works the same way. It doesn't. And if you're applying one blanket strategy across all of them, you're probably leaving visibility on the table—or worse, optimizing for the wrong thing entirely.

Once you understand how each type of LLM generates its responses, the right strategy becomes a lot clearer. In this chapter, we're going to look at the three main types of LLM responses and, more importantly, the kind of content each one rewards. The next three chapters will then dive into specific tactics for each response type.

How LLMs Generate Responses

Do you answer every question in exactly the same way? Of course not—and neither does AI. Some models generate responses using only information they already know, some reach for fresh information, and some pull in additional “experts” for deeper research and reasoning.

In this chapter we're talking about response types, not obsessing over individual models or brand names. That distinction matters. Consumer‐facing platforms switch between models for almost every query, and most of them are running at least three different models in parallel:

• A flagship, high‐performance general model with a large context window and lots of skills, running on the latest tech.
• A “lite” or “flash” model for quick, lightweight answers—less complex, but faster and cheaper.
• A “deep ...