Key takeaways
- ChatGPT search draws on web content and shows linked sources. Being allowed to crawl and having clear, citable pages is the controllable part.
- OpenAI documents separate crawlers for search and training. You choose access per bot in robots.txt.
- There is no verified trick that guarantees citation. Accurate, specific, well-structured pages with named sources are the durable approach.
- Measure referrals and brand mentions in your analytics rather than trusting tools that claim to track rankings in assistants.
01
What is ChatGPT SEO?
People now ask assistants questions they used to type into a search box. OpenAI introduced ChatGPT search so answers can draw on the web and link to the sources. "ChatGPT SEO" is shorthand for making your pages findable and worth citing in that setting.
It overlaps heavily with ordinary SEO. The pages that are easy to crawl, clear about what they say and backed by evidence are the pages any retrieval system has the best chance of using. What is new, in our view, is the access layer and the way answers are assembled from passages, not whole pages.
02
What can you control?
OpenAI documents its crawlers and how to allow or block them through robots.txt, with separate user agents for search and for training data. Decide which you want deliberately, because blocking the search crawler keeps you out of ChatGPT search answers (OpenAI says opted-out sites can still appear as navigational links), while the training decision is a separate business choice.
Beyond access, control the content. Open each section with the direct answer, name the entity and the facts plainly, cite your sources and keep key claims in text rather than inside images or scripts that a crawler may not read.
- Review robots.txt for OpenAI user agents and decide per bot
- Make key facts present in server-rendered text
- Lead each section with the answer
- Cite primary sources and date the page
- Keep brand and product names consistent across the web
03
What can nobody promise?
No one outside the provider can guarantee that a page will be cited, or reliably measure how often it is. Claims of secret prompts, schema hacks or paid placements in assistants should be treated sceptically. Google likewise says that for its AI features the same foundational SEO practices apply and no special optimisation is required.
The same caution applies to the claims you make. The FTC has said there is no AI exemption from the laws on the books, and its enforcement sweep targeted deceptive AI claims, which is a reason to keep your own marketing about what your AI work achieves truthful.
| Claim | Reality check |
|---|---|
| Guaranteed ChatGPT citation | No provider offers or documents this |
| Special markup unlocks assistants | Not documented; structured data helps general understanding |
| Rank trackers show assistant position | Answers vary by user and time; treat as a sample |
| Block training, stay in search | Possible via separate user agents; check current docs |
04
How do you measure it?
Look for referral traffic from chatgpt.com and similar assistant domains in your analytics and segment it from other sources; referrer labelling varies, so treat counts as approximate. Check whether those visitors reach key pages and convert. Test a fixed set of buyer questions monthly, record which sources appear and treat the log as a sample, not a ranking.
Pair this with branded search growth and direct enquiries that mention an assistant. Small numbers are normal; the signal is whether the pages being cited are the ones you would want a buyer to read.
Was this useful? Send it to someone who owns this problem.
Primary sources and further reading
Use the source material to validate details against your own context and current platform configuration.
- OpenAI: Introducing ChatGPT search
- OpenAI: Overview of OpenAI crawlers
- Google Search Central: AI features and your website
- FTC: FTC announces crackdown on deceptive AI claims and schemes
- Google Search Central: Robots.txt introduction and guide
This field note follows the XenGrowth editorial policy: primary sources where available, visible limitations, material review dates, and no invented first-hand experience.
Stay with the problem
Related field notes
Local Search OptimizationReview Management Is a Workflow, Not a Reply Button: Ask, Route, Reply, Escalate, Use
AI and answer enginesAI Search Optimization: What Moves Citations and What Is Folklore
SEO & GEOTopical Map: Build a Cluster Plan Search and AI Can Read, and Know What Topical Authority Is Not