ChatGPT

ChatGPT answers in two fundamentally different modes, and the difference decides whether you can influence it at all. Sometimes it responds from what it learned during training — knowledge fixed months earlier, with no link and no source. Other times it searches the live web and builds the answer from pages it just fetched. The same question can trigger either.

That split matters because the two modes have different levers. Nothing you publish today changes what the model already learned. But you can shape what it finds when it does search — through crawlable content, clear structure, and coverage on the third-party sites it tends to pull from.

Because it is the assistant most people have actually used, ChatGPT is often where a buyer's first impression of your category forms. It is also the platform where a brand can be named confidently with no citation attached, which makes it the hardest of the four to audit by hand and the easiest to be invisible on without noticing.

ChatGPT has the largest share of everyday assistant use, which makes it the platform where the most first impressions of your category are formed. It is also the one most likely to name brands without linking to them — so being present there produces influence you will never see in a referral report, and being absent costs you buyers you will never know existed.

Asked "what are the best AEO tools," ChatGPT may return three names in a plain paragraph with no links at all — answering from memory. Asked the same question on another day, it searches, cites four sources, and returns a different list. Neither answer is wrong; they came from different mechanisms. A brand present in one and missing from the other looks inconsistent until you understand why.