How to test whether ChatGPT knows your e-Residency company (and what to fix first)
Ask ChatGPT about your company and you will get one of three results: an accurate answer, a wrong answer, or a polite “I could not find information”. Most founders react to the second and third with panic and start searching for a trick. There is no trick. There is a way to find out which of the three you are in, why, and what to fix first.
This guide is a test plan for an e-Residency company that wants to know how it appears in ChatGPT and what it can realistically influence. It relies on what OpenAI and Google document, and it states plainly where nobody knows.
What you can and cannot control
OpenAI’s crawler documentation describes three user agents, and each has a different job:
| User agent | What it does | Why it matters to you |
|---|---|---|
| OAI-SearchBot | Surfaces websites in ChatGPT’s search features | If you block it, your site is not used in ChatGPT search answers, although it can still appear as a navigational link |
| GPTBot | Crawls content that may be used to train generative AI models | Blocking it is a training decision. It is a separate setting |
| ChatGPT-User | Visits a page when a user asks ChatGPT to, and is not used to crawl automatically | It is not used to decide whether content appears in search |
The documentation adds that each setting is independent, and that a change to robots.txt can take around 24 hours to be picked up for search. Those are the facts. What OpenAI does not publish is how it ranks or chooses sources for a given answer, so anyone who tells you exactly how to get cited is guessing.
Step zero: check that the door is open
Before testing anything, open your robots.txt (yourdomain.com/robots.txt) and look for a rule that blocks OAI-SearchBot, or a blanket rule that blocks every unknown bot. Many sites block AI crawlers by accident through a security plugin or a default setting.
If you want your site to be eligible for ChatGPT search while keeping it out of training, the two settings can be different. A simple example looks like this:
User-agent: OAI-SearchBot Allow: / User-agent: GPTBot Disallow: /
Whether to block GPTBot is a business decision, not an SEO one. The point is that the two are separate, so make each choice on purpose. Wait about a day after any change before you test.
Build a prompt set that reflects real buyers
Do not test by asking “what is [your company]?” only. That tells you whether the system knows your name, and nothing about whether a buyer would find you. Write fifteen prompts in three groups:
| Group | Example shape | What it tests |
|---|---|---|
| Brand | ”What does [company] do?” and “Is [company] a legitimate business?” | Whether your facts are correct |
| Category | ”Which companies offer [your service] for [your buyer]?” | Whether you are mentioned at all |
| Problem | ”How do I [problem you solve]?” | Whether your content is used as a source |
Run each prompt in a fresh conversation, log the date and the exact wording, and copy what the assistant says about you, including the sources it shows. Do it on a logged-out browser or a clean session if you can, so that your own history does not shape the answers.
How RankMaster runs this test for clients
RankMaster is a Brazilian SEO and GEO agency that works remotely and in English with e-Residency and Estonian companies. This test is part of how we work, inside three recurring deliverables:
- Monthly plan. A written plan for the month: which prompts we will track, which pages we will improve and which technical checks come first.
- Monthly report. The log of results compared with last month, what changed in Google Search Console and how the company appears in AI answers, including prompts where nothing moved.
- Daily monitoring. Checks for pages leaving the index, robots.txt changes and broken links that could cut the site off from crawlers.
We do not promise citations, because no one can. If you want to see how we apply this to an e-Residency company, read our guide to SEO for e-Residency companies, see about RankMaster or ask for a free analysis in the contact section.
Repeat it, because answers vary
AI answers change from one attempt to the next. Run each prompt at least three times on different days and treat the results as a range. A company that appears in two of three attempts is in a different position from one that appears in none, and a single lucky or unlucky attempt proves nothing. This is why the log matters more than any one screenshot.
Read the results and pick a fix
Match what you saw to the most likely cause. These are working hypotheses, not diagnoses, and the right response is to change one thing at a time.
| What you see | Likely cause | First fix |
|---|---|---|
| Wrong facts about you (address, services, founder) | Inconsistent or outdated information across sources | Update your About page, LinkedIn and directories so the same facts appear everywhere |
| ”I could not find information” | Thin public footprint or crawler access blocked | Check robots.txt, then publish a clear About page and service pages |
| Mentions competitors but not you | Your pages do not answer the category question | Write pages that answer the exact question buyers ask, in plain language |
| Mentions you with a source that is not your site | A third party describes you better than you do | Make your own page the clearest description of what you do |
| Mentions you correctly | Nothing to fix | Keep the pages current and keep testing |
Keep your entity facts consistent
For a company registered in Estonia, the registry entry is the official record of your name and registry code. The Estonian e-Business Register lets users query companies by name and registry code, so your name and code should be written the same way on your site. Put the legal name, registry code and contact details on one page, and use the same wording on LinkedIn and in every directory. If your brand name differs from your legal name, say so once and clearly.
Google’s own guide to AI features says no special markup or extra files are needed to appear in its AI search features. We apply the same principle here: start with clear pages, accurate facts and a crawlable site before anything exotic.
Common mistakes
- Testing once and drawing conclusions. One answer is an anecdote.
- Blocking every bot and then wondering why you are invisible. Check robots.txt first.
- Fixing five things at the same time. You will not know what worked.
- Paying for a “guaranteed citation”. Neither OpenAI nor Google documents a way to buy one.
- Letting facts drift. A new address or service that appears on one page and not on the others invites wrong answers.
Frequently asked questions
Can I make ChatGPT recommend my company?
No. OpenAI does not document a way to buy or guarantee a mention. You can make your site accessible and your facts clear, then measure.
Does blocking GPTBot hide me from ChatGPT search?
According to OpenAI’s documentation, the settings are independent. GPTBot relates to training, and OAI-SearchBot relates to search results.
How long does a robots.txt change take?
OpenAI says it can take around 24 hours for its systems to adjust for search results.
Do I need an llms.txt file?
Google states that its search ignores such files, and OpenAI’s crawler documentation does not mention them. It is optional and low cost, but it is not a substitute for clear pages.
Is there an official report for ChatGPT visibility?
Not that we know of. For Google, Search Console has a performance report for generative AI features. For ChatGPT, your own prompt log is the best evidence you have.
Your first test this week
Open your robots.txt and confirm that OAI-SearchBot is not blocked. Then write ten prompts, five about your brand and five about the problems you solve, and run each one in a fresh session. Save the results in a spreadsheet and pick the single most common problem. If you want help reading the results, ask for a free analysis through the RankMaster contact section.