How to write answers AI tools can quote

Put the direct answer in the first one or two sentences under a heading that matches the question. Make that passage understandable on its own, state facts plainly with names and numbers, then add the detail below. This is how people prefer to read too, and it gives search features and AI tools a clean passage to use. Nothing guarantees a citation, but vague or buried answers rarely get quoted.

Why this works

Answer engines and AI tools pick passages that clearly respond to a question. If your page makes them dig through a long introduction to find the point, a competitor's page that states it up front is easier to use. Google's own guidance for its AI search features points to the same basics: crawlable pages, original content and clear structure. You do not need special tricks.

Structured data can support clear writing, but it cannot replace it. If you are new to markup, structured data for beginners explains what to add first. And if you are wondering how this fits with ordinary search work, SEO, AEO and GEO: what is the difference sorts out the terms.

Seven steps

1. Use the question as the heading

Write a heading in the words people actually use, such as "How long should a title tag be?" Headings that match real queries make the page easier to match to the question.

Keep the levels in order too: one H1 for the page, H2s for the main questions, H3s for the parts inside them. If your headings are out of order or missing, how to fix your heading structure shows how to repair them.

2. Answer in the first sentence

Lead with the answer, then explain. Compare:

• Buried: "Title tags have been an important part of SEO for many years, and there are many opinions about the best approach to length..."
• Direct: "Aim for about 50 to 60 characters. Google has no official limit, but it cuts off titles that are too wide for the result."

The second can be quoted as it stands. The first cannot.

3. Make each answer stand alone

Avoid "as mentioned above", "this", "the previous section" or pronouns that need earlier context. If a sentence were pulled out and shown alone, it should still make sense. Name the subject: write "llms.txt does not block crawlers", not "it does not block them".

4. Be specific

Use exact names, numbers, units and dates. "Core Web Vitals thresholds include a Largest Contentful Paint within 2.5 seconds" beats "pages should load fast". Specific facts are easier to quote and easier to check.

5. Keep one idea per paragraph

Short, focused paragraphs are easier to lift cleanly and easier for people to scan. This does not mean chopping your content into fragments for machines. Google lists breaking content into tiny pieces as a tactic it says you can ignore. Write naturally and clearly, and let the structure follow the ideas.

6. Use lists and tables where they fit

Steps belong in a numbered list. Comparisons belong in a table. Definitions belong in a short sentence that starts with the term. Match the format to the content type, and do not force a list where prose reads better.

7. Show who wrote it and when

Add an author name, a publish date and an honest updated date. A simple line near the top of the page is enough:

Written by Jane Doe · Published 3 March 2026 · Updated 12 September 2026

Only change the updated date when you have really changed the content. State where your claims come from, and link to the primary source when you cite one. Readers and systems both use these signals to judge whether to trust a page. For the how-to, see how to add a publish and updated date to your pages.

A before and after

Before:

There are lots of things to think about when it comes to robots.txt and AI. Many site owners are unsure. It really depends on your goals, and there are several factors involved in deciding.

After:

Blocking GPTBot in robots.txt stops OpenAI's training crawler from reading your site. It does not remove you from Google Search, because Googlebot is a separate crawler. To block it, add a group with User-agent: GPTBot and Disallow: /.

The second version answers the question, names the exact crawler and rule, and could be quoted in full.

A full worked example

Here is a complete section, with the heading, the opening answer and the detail underneath, written the way the steps above describe.

How long should a title tag be?

Aim for about 50 to 60 characters, with the most important words first. Google has no official character limit, but it shortens titles that are too wide for the search result.

Width is measured in pixels, not characters, so a title full of wide letters such as W and M is cut off sooner than one full of narrow letters such as i and l. A commonly cited desktop width is around 600 pixels, which is a rule of thumb rather than a guarantee.

A title that is too long loses its ending, not its ranking. If your brand name sits at the end, put it last and keep it short, because it is the part you can afford to lose.

Notice what the first two sentences do. They answer the question, name the subject, give a number and state the limit. A reader who stops there has what they need, and so does a tool that lifts only that passage. Everything after it is detail for people who want more.

What to check on the technical side

A good answer is no use if a crawler cannot reach it.

• Allow the crawlers you want. Check your robots.txt for blocks on AI crawlers. The AI crawler checker shows which ones your rules allow.
• Put the text in the HTML. If the page is an empty shell that fills in after scripts run, some crawlers will see almost nothing. See the quick test below.
• Check firewalls and bot filters. Some security tools turn crawlers away even when robots.txt allows them.
• Confirm an AI tool can actually see you. How to check if ChatGPT can see your site walks through it.

A quick test: is your answer in the HTML?

Open your page, then view its source (in most desktop browsers, right-click and choose View Page Source). Search the source for the first sentence of an answer. If you find it, the text is in the HTML. If you do not, the text is probably added by scripts after the page loads, and a crawler that does not run scripts may never see it.

You can run the same test from a terminal. This command fetches the raw page and searches it for a word:

curl -s https://halfdev.com/quick-wins/how-to-fix-heading-structure | grep -i "heading"

If the command prints matching lines, the raw HTML contains that text. If it prints nothing for text you can plainly see in your browser, it is being added by scripts.

What not to do

• Do not stuff keywords or repeat the question mechanically.
• Do not rewrite everything in an unnatural, machine-first style.
• Do not invent statistics or quotes, or add fake mentions around the web.
• Do not promise yourself a citation. No outside tool can see why an AI system chose a source.

Check your page

Run your page through the free AEO checker to see how it looks on answer-engine signals, including whether AI crawlers can reach it. If you want to understand what those signals mean next to classic SEO, the SEO, AEO and GEO explainer is the place to start.

Then read your top three pages as if you were a stranger who only reads the first sentence under each heading. If that sentence does not answer the question, rewrite it first.

Related posts

• How to fix your heading structure
• How to add a publish and updated date to your pages
• How to add FAQPage schema, with copy-paste JSON-LD
• What FAQ schema does and does not do
• SEO, AEO and GEO: what is the difference
• Structured data for beginners: what to add first
• How to check if ChatGPT can see your site