Guide

How to appear in ChatGPT: what works and what does not

Check ChatGPT search access, crawler controls and cited sources. Use a repeatable study to decide what to work on without promised positions.

To assess why ChatGPT does or does not recommend your company, collect the answers to relevant buyer questions. For answers using web search, inspect cited sources and check whether your pages are eligible to appear. These checks inform a study; they do not guarantee a recommendation.

Request the study

ChatGPT can use web search when current information would help answer a question. Search-enabled responses may cite sources, as OpenAI explains in its search documentation. Inspect the actual responses rather than assuming every answer used search or that a search ranking guarantees a recommendation.

OAI-SearchBot vs GPTBot: search or training

OpenAI’s crawler documentation distinguishes two controls:

  • OAI-SearchBot is for search. Allowing it supports eligibility for ChatGPT search results. OpenAI says opted-out sites will not appear in search answers, although they can still appear as navigational links.
  • GPTBot is for training. Its setting is independent of OAI-SearchBot. Allowing search access does not require allowing training access.

Review crawl access and the sources cited in your category separately. Permission to crawl is not evidence that a brand is mentioned, recommended or cited. Those outcomes must be counted in the saved responses.

If you sell through retailers, pharmacies or distributors, there is a bigger trap: measuring the wrong instrument. Grading your website tells you about your website. Your buyers walk a path that may never touch it: they ask an assistant, open a retailer app, read two reviews, and pick up the product. Visits to your brand’s site on that path: zero. The category conversation is the frame that matters, not your domain.

What does not work

Google’s guidance below applies to generative AI features in Google Search. It does not establish what ChatGPT or other assistants use or require. Read Google’s guide to generative AI features in Search:

For Google Search, the guide says foundational SEO best practices remain relevant to generative AI features. That is Google’s guidance for its own Search products, not proof that every assistant works the same way.

What actually moves it

What is left is unglamorous and effective:

  • Inspect the sources actually cited. Candidate sources for consumer studies include retailer pages, reviews, consumer media and regulators. For B2B, inspect comparisons, industry media, documentation and relevant communities. The study identifies which sources appear for the chosen questions.
  • Check access to your own content. Review OAI-SearchBot access, page availability and the technical requirements relevant to each search surface. Access alone does not establish recommendation frequency.
  • Keep published facts consistent. Check the brand name, product details and attribution across the pages you control. Treat conflicting facts as a research finding rather than assuming they caused a particular answer.
  • Prioritize feasible source work. Use the study to decide which retailer content, editorial opportunities or other sources merit attention. Publishing or earning a mention does not guarantee that a model will use it.

None of this is llms.txt. All of it is the open web.

Why measure before you optimize

Repeating a category question can produce different answers. A single screenshot records one response; it does not establish how often your brand is mentioned or recommended. Save the responses, repeat the agreed questions and report the observed rates with their denominators.

This is also why generic advice stays generic. “Be present in the sources the models cite” is only actionable once you know which sources those are in your category, how often they appear, and who controls them. That is not an opinion. It is a count.

From generic advice to your category

The study makes this page specific to you:

  1. We build a prompt bank from your category and documented demand: the questions your buyers actually ask, unbranded, brand-named and competitive.
  2. We run it across AI models, with repetitions, and read every response.
  3. You get the ordered source list with owners: the retailer page you can rewrite this quarter, the earned media to pursue, the regulator page nobody displaces and you work around.
  4. Every action item above stops being a category and becomes a name.

The method, the deliverables and the sample reports are on the study page: /auditoria-visibilidad-ia/. And when the question stops being which sources move the answers and becomes who works them, the work on the sources is the service that executes that list: the retailer pages, the commissioned editorial, the earned mentions it names.

Start with the sources, not the hacks

There is no file that turns ChatGPT into a fan of your company. There is a source list, and it is measurable. Tell us your category and we will scope the study that names it.

Request the study

The evidence behind it

The devices below are the same ones a real report is built from. The numbers are an invented sample; the shape is not.

Device

The same question, asked three times

Why a single screenshot is an anecdote in both directions. Invented sample.

Run 1

  • Bloqsol
  • Dermalux

Run 2

  • Dermalux
  • Bloqsol
  • Solaria (listed, not recommended)

Run 3

  • Bloqsol
  • Dermalux

Same question, asked again. The lists move, and Solaria appears in one of them. A single screenshot is an anecdote, in both directions: false comfort and false panic. Screenshot the first answer and you conclude the brand is invisible; screenshot the second and you conclude it is doing fine.