Skip to content

The mechanism

How to appear in ChatGPT answers

Three things decide it, and only two are in your control. Knowing which is which stops a lot of wasted effort.

The short answer

An assistant names you when it can retrieve a source about you, when several sources agree on what you are, and when what they say matches the question being asked. You control the clarity of your own pages and much of what third parties say. You do not control the model's training, its retrieval choices, or when they change.

What ranked on 23 September 2026: This query returns tool vendors and agency guides, most listing tactics without distinguishing between what a business controls and what it does not. That distinction is the most useful thing to establish first.

The three conditions

Every tactic worth doing maps onto one of these. Anything that maps onto none of them is decoration.

  • Retrieval: the assistant has to be able to fetch and parse a page about you, which is ordinary technical accessibility work
  • Consensus: several sources the model trusts have to describe you the same way, which is why third-party accuracy matters more than your own adjectives
  • Match: the description has to fit the question, so a page about what you do for one kind of buyer wins that buyer's question and no other
  • Outside your control: training data, model updates, which sources a given assistant favours this month
  • Also outside your control: whether a given phrasing triggers live search at all, which varies by model and by question

Things that sound right and are not

  • ClaimAdd an llms.txt file and you will be cited.

    It is cheap and harmless and there is no published evidence it decides citations. Do it after the three conditions above, not instead of them.

  • ClaimMention your brand name more often.

    Repetition on your own site is not consensus. What moves this is other sources agreeing, which repetition cannot manufacture.

  • ClaimOne prompt did not mention us, so it is broken.

    These systems are non-deterministic. The same question can produce different sources within an hour, which is why a fixed prompt set sampled over time is the only sane measurement.

Where to start

  1. 01Check that your key pages can be fetched and read without running JavaScript, because a page that cannot be parsed cannot be quoted.
  2. 02Rewrite those pages so each one opens with a complete, self-contained answer to a single question.
  3. 03Find the three sources assistants currently cite for your category questions, and get your entry on each of them correct.
  4. 04Answer the real questions in public where your buyers ask them, under your own name.
  5. 05Sample a fixed prompt set monthly, and judge on the trend rather than on any single answer.

Questions

Does ChatGPT use Google's index?
Assistants vary in how they retrieve, and the arrangements change. Assume that being findable and clearly written helps across all of them, because every current implementation depends on retrieving something readable.
How long until a change shows up?
A page edit can be reflected within days of a recrawl. A change in how you are described more broadly takes longer, because it depends on other sources catching up.
Can we pay to be included?
Not in the organic answer today, and the platforms are testing advertising in these surfaces. Treat sponsored placement as a coming option rather than a current shortcut.

Related

Doing this yourself and want a second opinion on what the assistants say about you? Tell me what you are seeing, or read the AEO Playbook for the whole framework.