Blog Posts That Get Mentioned in ChatGPT

When you ask ChatGPT a question, the answer doesn’t always come from the same source. Sometimes it draws from its training data. Other times, it runs a live web search in the background. That second behavior, known as a query fan-out, is reshaping how content earns visibility inside AI-generated responses. And according to a recent test of 90 prompts, the type of content that triggers that fan-out is surprisingly narrow.

In the experiment, commercial prompts triggered web searches 78.3% of the time, while informational prompts did so just 3.1%. That gap is massive, and it points to a strategic shift for anyone writing content with AI visibility in mind. The old question was how to rank. The new question is which pages open the door to the fan-out in the first place.

The test covered three industries: beauty, legaltech and regtech, and IT. Of the 90 prompts, 65 were informational, 23 were commercial, and only one each were branded or transactional. Despite the heavy skew toward informational intent, 20 prompts triggered fan-out and 18 of those were commercial. That means 90% of all fan-out triggers came from commercial queries.

When the system did expand a prompt, it produced an average of 2.1 background searches per trigger. Across the 42 fan-out queries generated, 39 were commercial, two were branded, and only one was informational. Even when a broad informational prompt sparked expansion, the system often shifted toward evaluative, solution-seeking language. For example, “I need an open-source document management system. What can you suggest?” was classified as informational at the prompt level, but the downstream fan-out moved straight into solution recommendation.

This pattern aligns with a broader design trend in generative search. Google describes its AI Mode as breaking a question into subtopics, running parallel searches, and synthesizing results into a single response. ChatGPT appears to follow a similar logic. The implication for content strategy is clear: informational content alone is unlikely to align with fan-out paths, at least in this dataset.

The takeaway is not to abandon educational writing. Rather, it’s to build commercial bridges into top-of-funnel content. If you want visibility in AI answers tied to product selection, vendor discovery, or option narrowing, your content model needs pages that match those downstream commercial branches. That includes best-of lists, comparison pages, feature-led category explainers, alternatives pages, evaluation FAQs, and recommendation-oriented paragraphs embedded inside broader articles.

A purely educational piece that explains a category without naming products, tradeoffs, features, use cases, pricing logic, or selection criteria is much less likely to align with the fan-out paths observed here. The strategic shift is subtle but powerful: don’t just answer the obvious question. Anticipate the next evaluative step the system is likely to generate in the background.

This result is directional, not universal. Ninety prompts reveal a pattern, not a stable law of AI retrieval behavior. The sample is uneven, with branded and transactional prompts barely represented. Some industries may be easier to express in product-discovery language. And this is an observational analysis of ChatGPT, not a controlled test of Google AI Mode. Still, the pattern is strong enough to warrant a closer look.

The next step is to map triggered fan-outs back to specific content formats. The goal isn’t just to confirm that commercial intent wins. It’s to identify which page templates and passage structures best cover the fan-out branches AI systems prefer.