Want to get your business cited in AI search answers? Start with the questions your customers are asking. Want to avoid cognitive surrender? Don’t outsource your own critical thinking. As content strategies increasingly focus on writing for machines, customer needs and original thinking risk getting lost. I argue that relevance matters for citations, but developing our own ideas helps preserve the reasoning skills we're increasingly tempted to outsource.
Within a website, pages that matched the questions people ask were cited more. FAQs, structured data and summary boxes weren't associated with higher citation rates between pages on the same site.
Genre can be a gateway into the customer's mindset, translating needs into usable content.
No study shows that machines reward original ideas, and models tend to produce similar ones when used for ideation. Developing and testing our own perspectives can counter homogeneity, and may also help preserve our own critical-thinking skills.
Does formatting for answer engine optimization (AEO) get pages cited?
Not by itself, according to a recent October 2026 preprint on AI citations. Analyzing more than 158,000 AI-generated answers and 468,000 citations from ChatGPT, Gemini and Perplexity, researchers found that pages matching users' questions were cited more often within the same website. Conversely, FAQ blocks, structured data and summary boxes weren't associated with higher citation rates (in those within-site comparisons).
These findings suggest that writers should focus less exclusively on AEO formatting and more on addressing customer needs; that’s where genre can be put to use.
Does writing to genre help get citations?
The impact of writing to genre hasn't been tested, but it can be a useful forcing function to get writers to think about the needs of their readers. In the study on AI citations, ChatGPT wrote new search queries for each answer, reinforcing why exact keyword matching isn't enough in the age of AEO. Instead, writers have to address the meaning behind queries: their customer’s underlying intentions and needs.
Genre can be a route to getting into that customer mindset. It’s the structural framework for your writing: the shared conventions your audience expects, and it dictates the content your assets need to contain:
| Genre type | The reader's need | What the page has to contain |
|---|---|---|
| Tutorial | Do something and get a working result | Prerequisites, an executable process, the expected result, where it commonly goes wrong |
| Explainer | Understand why something works | Definitions, relationships between concepts, mechanisms |
| Visionary thought leadership | See an assumption challenged | The assumption, where it breaks, a proposed new framework |
| Practitioner thought leadership | Evaluate whether an idea survives in a real-world setting | Decisions, constraints, trade-offs, outcomes |
Let's look at a technical tutorial from the stance of genre and AEO. A tutorial that explains every concept beautifully but never gives the reader a working procedure fails its genre, even if its headings and schema are clear.
In other words, markup tells a machine how the page is organized, but genre tells the writer what it has to contain.
If I'm building writing skills for agents, I start by defining genre expectations for each asset type. This improves the direction given to the agent while requiring the human to consider what the reader actually needs.
Do machines reward original ideas?
There's no clear evidence that originality earns more AI citations. In fact, a study of Google's AI Overviews found that more predictable passages were more likely to be cited.
Research also raises concerns about the diversity of AI-generated ideas. A study comparing 22 LLMs and 102 humans found that LLM responses to creative prompts were substantially more similar to one another than human responses. In a separate 36-person study, ChatGPT users generated more similar ideas than participants using another creativity tool.
Together, these findings suggest that relying on AI for ideation risks making our content more homogeneous, in a landscape where there’s no obvious reward for originality.
If relevance matters more than originality for citations, why say something unique?
Relevance might bring people to your page, but novelty can give them a reason to read it. Unique perspectives can also differentiate a brand. Yet there's another reason to develop our own ideas: protecting our own ability to think critically.
Researchers at Wharton coined the term cognitive surrender: accepting AI outputs with minimal scrutiny rather than constructing our own answers. In a study of 1,372 participants, researchers found that participants accepted incorrect AI answers nearly three-quarters of the time. AI-assisted participants also showed greater confidence overall.
A 2025 Microsoft Research and Carnegie Mellon survey of 319 knowledge workers found a related pattern. Greater confidence in AI was associated with less reported critical thinking, while confidence in one's own abilities, including evaluating AI outputs, was associated with more.
The implication isn't that we should stop using AI. It's that we need to be aware of and deliberate about what we choose to outsource.
Sure, citation research encourages us to think about customer relevance, and genre can help us get there. But I'd argue that the more important work isn’t validated by any AEO studies or citation results. Instead, it’s evidenced in our own mental capacity: in our continued ability to form our own viewpoints, test them against evidence, and decide what we actually think.
Because the goal isn't simply to produce content AI can find. It's to keep doing the critical thinking that makes content worth finding, and that protects our own capacity to reason in the age of AI.
