10 things to put in your article if you want ChatGPT to actually cite it

"How do I rank in ChatGPT?" is the question everyone asks, and it's slightly the wrong one. ChatGPT doesn't rank pages the way Google does - there's no results page, no position one. What it does is pick sources it trusts enough to quote or summarise when it answers a question. Getting picked comes down to specific, checkable things you can put in an article. Here are ten of them, each backed by actual research rather than a guess.

1. Answer the question in the first couple of sentences

AI tools extract from the beginning of a section outward, working through the page in order. If a section opens with background or scene-setting instead of the answer, the model may never reach the actual answer, or it lifts the context and treats that as the answer instead. Profound's content guide walks through this mechanism directly and calls it the single most effective structural change you can make to a page. Lead with the answer, then explain it.

2. Use real numbers, not vague claims

"Many businesses struggle with this" tells a model nothing it can quote. A study from Princeton and IIT Delhi tested this directly and found that adding specific statistics to a page increased its visibility in AI-generated answers by up to 34%. Specific beats vague, every time.

3. Quote people and name your sources

The same study found that adding quotations from credible sources lifted visibility by up to 40%, and simply citing a source added 27%. A model is more confident quoting something that's already attributed to someone real.

4. Write real FAQ content, not just FAQ schema

FAQ schema markup used to be treated as a shortcut into AI answers. It isn't any more: Google dropped FAQ rich results from search entirely in May 2026, and Ahrefs tested 1,885 pages that added FAQ schema and found the markup itself made no meaningful difference to citation rates. What still works is plainly written, visible question-and-answer text on the page - every major model reads that regardless of the code underneath it.

5. Add a comparison where one genuinely fits

Seer Interactive's 2026 study of 7,683 pages carrying over 47,000 AI citations found comparison and review-style content among the most consistently updated - and most consistently cited - formats it tracked. If your article naturally involves comparing two options, build it as an actual comparison rather than burying the difference in a paragraph.

6. Check that AI crawlers can actually reach your page

None of this matters if the bot can't get in. BuzzStream checked robots.txt files across major publishers and found PerplexityBot blocked on 67% of the sites they looked at. Often a leftover default setting nobody had revisited, not a deliberate choice. Check your own site allows GPTBot, PerplexityBot and ClaudeBot before you worry about anything else on this list.

7. Keep your claims the same across your other pages

A single great article rarely holds a citation on its own. A 90-day study across ChatGPT, Claude and Perplexity found citation half-lives of roughly 3 to 9 weeks depending on the platform, with evergreen content holding a citation around twice as long as one-off posts. Repeating the same specific claims across several pieces, rather than saying something slightly different every time, is what keeps a citation alive.

8. Update old pages, not just publish new ones

The same Seer Interactive study found that 75% of pages cited by AI models had been updated within the past year - but more than a quarter of those "fresh" pages were originally published over two years earlier. The freshness came from maintenance, not from writing something new. An old article you keep current can out-cite a brand new one you never touch again.

9. Get mentioned somewhere other than your own website

Ahrefs analysed 75,000 brands and found branded web mentions correlate at 0.66 to 0.71 with AI visibility. Backlinks, by contrast, correlate at just 0.19 to 0.29. Writing about yourself on your own site only goes so far; being talked about elsewhere is what the data actually rewards.

10. Include something original - a number, a result, a real example

Fractl surveyed 150 marketers in 2026 and found only 15% currently prioritise original research or proprietary data as a strategy, despite Fractl's own analysis calling it one of the hardest things for AI to simply replicate or summarise away. A real result, a real number from your own work, or a genuine example beats another summary of what's already been written a hundred times.

Putting it together

None of these ten things work in isolation, and doing one of them brilliantly doesn't make up for ignoring the rest. For how the models actually build a shortlist once your content clears these basics, see How marketing agencies get found in ChatGPT and Perplexity.

A few common follow-up questions

Can I do all ten in one article?

You can aim for most of them in a single piece - a direct opening answer, a real statistic, a named source, and visible FAQ content all fit naturally in one article. Freshness, cross-page consistency and off-site mentions build up over several pieces and over time, not in one sitting.

Which one matters most if I can only fix one thing?

Crawlability. A brilliantly written page a bot can't reach doesn't get a chance to do anything on this list.

Does this apply to Perplexity and Google's AI Overviews too, or just ChatGPT?

All ten points apply across the major platforms - the studies above cover ChatGPT, Perplexity, Claude and Gemini, not just one. Citation rates and behaviour vary by platform, but what gets an article picked up in the first place doesn't change much between them.