claude and perplexity read your whole page. chatgpt skims your domain. write for both.
not every AI platform decides who to recommend the same way, and that difference should change how you structure your site, not just what you say about yourself.
Domain-level trust is defined as an AI system treating a whole website as a credible source for a topic and routing a visitor to it without selecting one exact page. Page-level selection is defined as the system choosing one specific URL as the source. ChatGPT and Gemini lean toward the first pattern; Claude and Perplexity toward the second.
the split, plainly
ChatGPT and Gemini tend to trust a domain broadly and route a visitor somewhere on it, rather than pinpointing one exact page with certainty. Claude and Perplexity behave differently. They tend to pick specific pages and lean toward long-form, well-sourced content when deciding what to cite.
Neither approach is better. They're solving the retrieval problem differently, and it means a single, shallow bio page might be enough to get you domain-level trust from one platform while giving another platform nothing specific enough to actually cite.
| platform | tends to | what that rewards |
|---|---|---|
| ChatGPT | trust the domain, route broadly | a clear, consistent overall footprint across many pages |
| Gemini | similar domain-level trust pattern | same as above |
| Claude | pick specific pages, favor long-form | one page that goes deep on a specific, answerable question |
| Perplexity | pick specific pages, favor long-form | same as Claude |
why this matters more than it sounds like it should
If your entire online footprint is a single bio page and a handful of social captions, you might get picked up by ChatGPT or Gemini as a broadly trusted name in your niche, since domain-level trust doesn't require deep individual pages. But you're giving Claude and Perplexity almost nothing to work with, because there's no single page detailed enough to be the specific thing they'd cite.
The reverse is also true. A creator or small operator who publishes one genuinely deep, well-sourced piece, a real breakdown of your process, your rates, your niche, with actual specifics, gives Claude and Perplexity exactly what they're looking for, and that same piece still reinforces the broader domain trust ChatGPT and Gemini are weighing.
what to actually do with this
You don't need a different site for each platform. You need at least a few pages that go genuinely deep, not just a scattering of shallow ones, alongside the broader consistent footprint that domain-trust platforms reward.
- Have at least one page per core topic that goes long and specific. Not a paragraph, a real, sourced, detailed answer to one question your audience actually asks. This is the page Claude and Perplexity have something to cite.
- Keep your one-sentence positioning consistent across every page, not just the deep ones. This is what builds the broader domain trust ChatGPT and Gemini are weighing.
- Don't sacrifice depth for volume. A dozen shallow pages restating the same shallow bio in different words gives every platform less to work with than three genuinely deep ones.
How a deep page actually gets read once a platform retrieves it, split into passages, scored section by section, is its own mechanic, covered in how ai chunks your page. Structure the deep pages with that in mind and both patterns get served by the same writing.
the practical takeaway
If you've been optimizing purely for "getting mentioned anywhere," it's worth checking whether you have at least a few pages substantial enough for a platform that reads the whole thing, not just the ones that skim a domain and move on. Both matter. Most creators and small operators, without realizing it, have only built for one.
This same principle underlies why your own site's internal search matters too, once a platform does send someone your way. internal site search and ai traffic covers what happens after the click.
give every platform something real to work with
the ai builder toolkit is the plain-text, structured setup behind this whole approach: the skills and templates that turn one clear positioning into deep, consistent pages a domain-skimmer and a page-reader can both use.
get the toolkitfrequently asked questions
Do different AI platforms cite content differently?
Yes. ChatGPT and Gemini tend to trust a domain broadly and route visitors somewhere on it. Claude and Perplexity more often select a specific page and favor long-form, well-sourced content.
Should I write different content for each AI platform?
No. You need a consistent footprint across many pages, which serves the domain-trust platforms, plus a few pages that go genuinely deep on one question, which serves the page-selection platforms. One site, two properties.
Why does Perplexity favor long-form content?
It tends to select specific pages that cover a full topic rather than one angle, so a thorough piece frequently outperforms several shallow ones on related subtopics.
How many deep pages do I need?
At least one per core topic you want to be known for. Three genuinely deep, well-sourced pages give every platform more to work with than a dozen shallow pages restating the same summary.
Does this change how I structure a single page?
Not fundamentally. Answer-first sections, short paragraphs, and specific sourced claims serve both patterns. The difference is in how many pages you need and how deep each one goes.
written by elisabeth hitz, certified in anthropic's ai fluency program (framework & foundations, and ai capabilities & limitations), plus claude 101 and claude cowork. published 26 july 2026.