Getting Cited: On-Page Tactics

Getting cited is mostly a writing problem, not a technical one. The engine reads passages and lifts the ones it can repeat cleanly. So write the sentence you want it to say back, put it where it can be found, and back it with evidence the engine trusts. This chapter is the on-page craft of that.

How do you write a passage an AI will repeat?

Write the sentence you want the AI to quote, then make it survive being lifted out. A quotable passage is a short unit that answers one question completely on its own. If you can cut a paragraph out of the page and it is still true and still clear with no surrounding context, it survives the chunking that retrieval does to your content. If it only makes sense in place, it gets read and discarded.

Three habits do most of the work. Lead each section with a direct answer of one or two sentences, before any setup. Name the entity plainly instead of writing "our tool" or "the platform," because the engine needs a subject it can attribute. And put the key fact, number, or definition in the first sentence or two of the chunk, not buried three paragraphs down where retrieval may never reach it. Keep answer paragraphs short, roughly four sentences, so the unit stays liftable.

What structure gets quoted verbatim?

Structure tells the engine where the answers are. Use question-shaped H2 and H3 headings that mirror how people actually ask, then answer immediately in plain text under each one. Headings do double duty here: they help readers and screen readers, and they mark the boundaries the engine uses to chunk your page. Headings and structure →

A few formats get quoted almost as they are written. Tables get lifted whole for comparisons, so put "X versus Y" content in a real HTML table, not a picture of one. Numbered lists get quoted for processes and steps. A short TL;DR near the top gives the engine a ready-made summary to grab. And all of it has to live in HTML text, not inside images or content that only appears after JavaScript runs, because AI crawlers render unreliably and cannot read what is not there. This is the same instinct that wins featured snippets in classic search, where Google extracts a clean answer near a matching heading. Featured snippets and PAA →

Does adding evidence really raise your odds?

Answer engines preferentially cite sources they can trust, so evidence is not decoration. Original data, a named and credentialed author, and citations to primary sources all signal that a passage is worth repeating and attributing. This is the same E-E-A-T judgment that classic search applies, carried into the answer. Content quality and E-E-A-T →

There is one study worth naming honestly. The GEO study found that adding citations, direct quotations from authorities, and statistics raised how often content appeared in generative answers, and that this helped more than keyword stuffing did. Reported gains land in the range of roughly 30 to 40 percent in that work. Treat it as one study pointing in a sensible direction, not a guarantee or a settled number. The underlying logic is sturdy even where the exact figure is not: give the engine a concrete, sourced claim and you give it something safe to cite.

The practical version is to make every strong claim traceable. Put a real name and credentials on the byline, so the author is a person the engine can weigh rather than an anonymous page. Link the primary source behind each number instead of a secondhand summary. And where you can, publish original data of your own, because a statistic that exists nowhere else gives an engine a reason to cite you specifically rather than a competitor who merely repeats the same industry figure.

The passage that gets cited is usually the one that made the engine's job easy: a plain answer, named plainly, with a number and a source attached. Vague, unattributed prose reads fine to a human and disappears in retrieval.

The on-page AEO checklist

Pulled together, the on-page work is a short list you can run against any page before you publish.

• 1. Front-load the answer. Open each section with a self-contained, one-sentence-or-two direct answer that names the entity.
• 2. Use question-shaped headings. Write H2 and H3 as the questions people ask, with the answer immediately below.
• 3. Add a TL;DR. Give a short summary near the top for the engine to grab.
• 4. Use tables and numbered lists. Put comparisons in tables and processes in numbered steps, in real HTML.
• 5. Attach evidence. Add original data, one concrete cited statistic per key claim, and a named author with credentials.
• 6. Keep it in text. No answers locked inside images or JavaScript-only rendering.
• 7. Keep chunks tight. Short answer paragraphs that stay true when lifted out of context.

On-page craft gets a passage quoted, but the engine also weighs whether it can identify who you are at all. That is the entity work, and it is where citations become durable rather than occasional. Entity and brand authority →