Content Structure: Question Headings, Answer First, Summary — What Is Known About AI

This guide shows how to structure an article so readers find the answer quickly: question headings, the answer in the first sentence, a short summary at the top and linked sources. These are writing habits for people. Where a study shows an effect on AI assistants, we say so — and where there is none, we say that too.

What is good content structure?

Good content structure means a reader can tell from each heading which question the section answers, finds the answer at the start of the section, and gets an overview of the whole page near the top. Google asks a similar question in its self-assessment for helpful content: “Does the main heading or page title provide a descriptive, helpful summary of the content?”

Why does content structure matter — for readers, Google and AI?

• Readers: Clear headings show readers where their answer is, and an early answer saves them time.
• Google: Google describes people-first content as “content that's created primarily for people, and not to manipulate search engine rankings.” Its self-assessment questions include descriptive headings and whether the content presents information “in a way that makes you want to trust it, such as clear sourcing”.
• ChatGPT and Claude: When they fetch a page live, they get the visible text, not the JSON-LD (Ramos study). Pages with question phrasing were found and cited more often — according to the author, because of the wording in the text.
• Sources, quotes, figures: In the GEO paper, they increased visibility in AI-generated answers by up to 40% in a test environment.
• Open: How individual assistants select passages, and whether summary blocks or answer-first paragraphs make a difference, has not been studied. We do not know.

Technique 1 — Question headings

Phrase your H2 and H3 headings as the question a reader would ask. Instead of “Our Pricing Model”, write “How much does [product] cost?”. Readers see at once whether the section answers their question. For this habit, the Ramos study also found an effect on AI assistants.

Step-by-step:

1. List the questions your readers actually ask about this topic.
2. Use each question verbatim (or near-verbatim) as an H2 or H3.
3. Keep headings short enough to read at a glance.
4. Avoid stacking two question headings without a paragraph in between.

Technique 2 — Answer first

Put the direct answer in the first sentence after each heading. Add context, caveats and sources afterwards. Readers who only need the answer can stop there; everyone else reads on. Whether this changes how AI systems use the page has not been studied — it is good writing either way.

Step-by-step:

1. Write your answer as one short sentence.
2. Place that sentence immediately after the heading, before any caveats.
3. Follow with explanation, context and a source link.
4. Avoid starting with “It depends…” or “There are many factors…” — that makes readers search for the answer.

Technique 3 — Summary at the top

For longer articles, add a labelled summary near the top. It tells readers in a few lines what the page covers and whether it is worth reading on. We know of no study showing that AI systems treat blocks labelled “TL;DR” or “Key takeaways” differently from other text.

Step-by-step:

1. Write a few bullet points with the most important facts of the article.
2. Place the block within the first screenful of content, below the lead paragraph.
3. Label the block clearly: <strong>TL;DR:</strong> or <h2>Key takeaways</h2>.
4. Keep each bullet to one sentence, and make sure it matches the text below.

Technique 4 — Link sources, quote, use figures

Back up claims with a link to the source, a quotation or a figure. Readers can check what you say. In the GEO paper, citations, quotations and statistics increased visibility in AI-generated answers by up to 40% in a test environment. More on this in the guide Citations & Source Links.

Code example — well-structured article section

The following HTML combines the techniques: a summary at the top, a question heading, an answer-first paragraph and a source link.

<!-- Summary block near the top of the article -->
<div class="tldr">
  <strong>TL;DR:</strong>
  <ul>
    <li>Question headings tell readers what each section answers.</li>
    <li>Put the answer in the first sentence after each heading.</li>
    <li>Link the sources for your claims.</li>
  </ul>
</div>

<!-- Question heading -->
<h2>Why should headings be phrased as questions?</h2>

<!-- Answer-first paragraph with a source -->
<p>
  Question headings show readers at a glance which question a
  section answers. Google's self-assessment for helpful content
  asks whether the main heading gives a descriptive, helpful
  summary of the content
  (<a href="https://developers.google.com/search/docs/fundamentals/creating-helpful-content">source</a>).
</p>

Common mistakes to avoid

• Vague headings: “More information” or “Overview” do not tell readers which question the section answers.
• Buried answers: Starting with background before the actual answer makes readers search for it.
• Summary that does not match the text: If the summary says something the article does not, readers lose trust.
• Overly long first sentences: An answer packed with caveats is hard to grasp. Keep the first sentence short and move the details to the next ones.
• Main questions hidden in H4/H5: Deep heading levels are easy to overlook. Put the main questions in H2 and H3.
• Answers only in the schema, not visible on the page: ChatGPT and Claude strip JSON-LD when they fetch a page live. Anything that exists only in the markup does not reach them — or your readers.
• Markup that differs from the visible text: When FAQ markup and visible text contradicted each other, Gemini Pro took the outdated FAQ answer 8 out of 20 times in the Ramos study. If you use markup, keep it identical to the visible text.

Official sources

• Rick Ramos (October 14, 2026) — study on schema markup and AI assistants (serptwerker.com)
• Aggarwal et al. — GEO: Generative Engine Optimization (KDD 2024)
• Google — Creating helpful, reliable, people-first content