Call · 30 min
MarketingMattia EspositoUpdated 10 November 202614 min read

How to rank in Grok. There are two doors, your X profile and your website, and xAI documents only one of them.

Grok decides for itself whether to search X posts or the web, and for websites xAI does not publish the name of its crawler. Here is what you can do on both sides, with a ten-point checklist in Word and PDF.

In brief

To show up in Grok you need a public X profile that states your trade and your town, and a website readable by a client that does not identify itself. Grok can search public X posts and run a real-time web search, and it makes the call itself, question by question.

The X side is documented. The X Help Center says posts from a protected account are not surfaced in answers; the xAI documentation says Grok also searches profiles and threads.

The web side is not. xAI names neither the search engine nor the crawler, and the word robots.txt does not appear anywhere in its documentation.

The ten-point checklist can be downloaded below, in Word and PDF, with no email required.

Written by: Mattia Esposito, AI and marketing consultant, founder of Itria AI. From Puglia, based in Turin since 2023, he works remotely in Italy and abroad. LinkedIn profile.

The “Grok presence” checklist, to download

A two-sided sheet, X and website, with ten checks, where to look and space to record the test you ran on Grok, in Word to fill in and PDF to print. You need access to the company’s X profile and to the site’s control panel. It takes twenty minutes.

The ten-point checklist for showing up in Grok, split between the X profile and the website, with the table of the sources Grok draws its answers from
FileContentsLink
Wordto fill in, in Italian

Grok’s sources, the ten checks split across X, website and Grokipedia, and the fields for the test.

come-comparire-su-grok-scheda.docx

PDFto print, in Italian

The same checklist, to tick off with a pen.

come-comparire-su-grok-scheda.pdf

N.What to checkWhere to look
1

The company’s X account is public, with posts left unprotected.

X, Privacy and safety

2

Name and bio state your trade and your town in a customer’s words.

The profile

3

The last ten posts talk about the work in the terms a customer would search for.

The profile

4

Who mentions the company on X: mentions, replies, threads.

X search

5

The pages you want cited open without blocking unknown clients.

CDN or firewall control panel

6

robots.txt has no rules for bot names copied from third-party lists.

yoursite.com/robots.txt

7

Copy, prices and contact details sit in the HTML, and the facts carry a source.

Page source (Ctrl+U)

8

Name, address and trade are written the same way on the site and on the X profile.

About page, footer, profile

9

Grokipedia has an article on the company or the sector, or one has been suggested.

grokipedia.com

10

A customer’s question, put to Grok on X and on grok.com, with the sources cited and the date.

Grok, once a month

On 7 August a friend sent me what ChatGPT was saying about me: “I can’t find a Mattia Esposito linked to itria.io as founder.” I had already written the llms.txt file and the structured data, the things people usually recommend. As far as AI was concerned, I did not exist.

Since then I do for the assistants what their own documents say, and I treat everything else as a hypothesis. With Grok the gap shows more than anywhere else. On the X side the rules are written down; on the website side xAI publishes almost nothing, and plenty of guides fill the gap with bot names xAI has never confirmed.

In September I checked the websites of 142 Italian companies that also sell abroad: twenty-eight out of a hundred and forty-three had the file explaining the company to an assistant, and in twenty-four a plugin had written it. In the post I put it like this: “An assistant doesn’t browse a site. It reads it, and reports what it finds written there.” Grok included. The full post.

Where Grok gets its answers

Grok answers with two searches, across public X posts and across the web in real time, and decides for itself whether to run them. The X Help Center puts it this way: “Grok has a unique feature that allows it to decide whether or not to search” public posts and the web. No document read sets an order between the two.

The xAI developer documentation gives the detail of both tools. With X Search, Grok runs keyword search, semantic search, user search and thread fetch. With Web Search it searches the web and opens pages, and the answer comes back with a list of sources. These are API documents: they describe what the model can do, not how the app uses it.

How Grok reaches an answer: it receives the question, decides whether to search public X posts, the web or both, and answers with its sources The questionon X or on grok.com Grok decideswhether and where to search X searchpublic posts, profiles, threadschecklist points 1-4 Web searchpages opened in real timechecklist points 5-8 The answerwith its sources
The two searches as described by X and xAI. No order between them is declared, and Grok makes the choice, question by question.

For a business, this means two doors to look after, with different rules. The first is the X profile, where X writes the rules. The second is the website, where the general rules apply, as the sister pages explain for ChatGPT and Perplexity, with one difference: for Grok there is no name to look for in robots.txt.

How to show up in Grok with your X profile

A public X account is the condition for the company’s posts to end up in a Grok answer. X says that protecting your account prevents your posts from being used to train Grok “or to be surfaced in response to user queries”. A protected profile is out by documented choice.

The second point is the profile itself. User search in X Search means Grok can find an account, and its posts too. Name and bio become a line answering the question “who does this job in this town”: the trade spelt out in full and the area, in the words a customer would use in place of the slogan.

Then the posts. Keyword and semantic search find posts about something; a profile that only publishes greetings and event announcements says nothing about its trade. Ten posts explaining a job done or answering a question customers often ask, with names and numbers, give Grok something to find.

The fourth point is out of your hands: who mentions you. Grok fetches threads and searches by user, so a mention on X from a customer or a supplier is material it can read. No xAI document says how much mentions count against the company’s own posts, or that Grok rewards or discards viral posts.

Grok searches the web: what we know and what xAI leaves out

xAI documents that Grok searches the web in real time and opens pages, but does not say which search engine it uses or which crawler reaches websites. The documentation is clear on the what: “The Web Search tool enables Grok to search the web in real-time and browse web pages to find information.” On the how, there is nothing more.

On the API side, a developer building an application can restrict the search to at most five domains, or exclude up to five. In the documentation’s example the domain used is grokipedia.com. The filter is for developers: people using Grok in the app neither see it nor control it.

For your site, the rules that apply with every assistant still hold. The copy has to sit in the HTML, because a crawler that does not run JavaScript cannot see what appears afterwards: the detail is in the guide to JavaScript and AI crawlers. The facts need a source written on the page, because Grok returns its sources, and a page with no verifiable facts gives a citation little to work with.

Grok’s crawler and robots.txt: what xAI says

Among the assistants covered by the guides on this site, xAI is the only one without a page about its own crawlers. On 10 November 2026, in the full xAI documentation, a file of about 2 MB, the words “robots.txt” and “crawler” do not appear, and no bot name does either. There is no bot page on x.ai, on docs.x.ai or in the X Help Center.

AssistantOfficial crawler pageName to use in robots.txt
Grok (xAI)

None, as of 10 November 2026.

Not declared

ChatGPT (OpenAI)

Overview of OpenAI Crawlers, on developers.openai.com.

OAI-SearchBot

Perplexity

Perplexity Crawlers, on docs.perplexity.ai.

PerplexityBot

Claude (Anthropic)

The crawler page in the Claude help centre.

Claude-SearchBot

Copilot (Microsoft)

Copilot searches with Bing, so Bing’s crawler is what counts.

bingbot

Third-party directories fill the gap with names that differ and contradict each other: GrokBot, xAI-SearchBot, xAI-Grok, Grok-DeepSearch, GrokAgent. None of them comes from xAI. A robots.txt rule for one of these names may do nothing at all if the name is wrong. The right names for the other assistants are in the guide to robots.txt for AI crawlers.

The x.ai robots.txt opens everything except one folder and declares a Content-Signal line with ai-train=yes, search=yes, ai-input=yes. It concerns the xAI website, and says nothing about how its crawler behaves on other people’s sites.

So what can you do in practice? If your server logs show a name containing Grok or xAI, note it down and do not write rules on trust: with no list of addresses published by xAI, anyone can put that name in the user-agent field. And if a firewall blocks every client that does not identify itself, the pages you want cited may stay closed to Grok as well.

Grokipedia: what it is and how to suggest an article

Grokipedia is an encyclopaedia written by Grok, launched on 27 October 2025, where readers propose edits and Grok checks them against sources and approves them. On 10 November 2026 the home page, at version 0.3, counted 1,263,052 approved edits. Every article ends with its sources.

Anyone with an xAI account can also suggest a new article. The Suggest page describes it in one line: name a topic, and Grok researches it, writes the article and checks it against sources. Which companies get an article is not written down anywhere: Grokipedia has no public rules page, and the /about address returns an error.

Two things should be kept apart. First, no xAI document says Grok uses Grokipedia as a source when it answers. Second, a study by Harold Triedman and Alexios Mantzarlis, published on arXiv in November 2025, found articles much longer than Wikipedia’s but largely derived from it, with far more citations to sources the Wikipedia community considers unreliable.

For a business this means one thing: if an article on the company or the sector exists, read it, because the sources it cites are what someone found about you. Wikipedia and Wikidata entries follow different, written rules, collected in the guide to Wikipedia, Wikidata and AI.

Grok in Italy and Europe: availability and the EU inquiries

Grok is available in every country where X is, so in Italy too, on the iOS and Android apps and on the web. For the data of X’s European users the lead authority is the Irish Data Protection Commission, which since 2025 has opened two inquiries involving Grok.

WhereWhat applies to GrokDate
Italy

Grok is available, because X is. The Irish authority leads for X; no decision by the Italian data protection authority on Grok turned up in this research.

X Help, copy of 1 November 2026

EU and EEA

The DPC is examining the use of European users’ public posts to train Grok. No decision found.

Inquiry opened on 11 April 2025

EU and EEA

The DPC is investigating intimate or sexualised images generated with Grok on X, including of children, under Articles 5, 6, 25 and 35 of the GDPR.

Inquiry opened on 17 February 2026

United Kingdom

Grok is available; for users’ data the authority is the ICO.

X Help, copy of 1 November 2026

United States

Grok is available; the setting for opting out of training is the same as in Europe.

X Help, copy of 1 November 2026

The first inquiry closely concerns anyone posting on X. The DPC says its aim is “to determine whether this personal data was lawfully processed in order to train the Grok LLMs”. The setting for opting out of training is under Privacy and safety, Data sharing and personalisation, Grok and third-party collaborators.

The setting covers training: the X Help Center links exclusion from answers only to the protected account. A business can therefore refuse the use of its posts for model training without locking its account. The second inquiry matters less for visibility and more for the reputation of the channel, and belongs in the sums when you decide how much to invest in X.

Can a business without an X account show up in Grok?

Yes, through the web: the company’s site and other people’s pages about it, if Grok decides to search there. What is missing is the best documented door, your own posts. For many small Italian businesses X is not the main channel, and the sums should be done on the time a profile takes.

A profile opened and left empty adds nothing to find. If the business has nothing to publish regularly, the effort pays back more on the website, which also serves ChatGPT, Copilot, Claude and Perplexity. If its customers or its sector do talk on X, a profile with a well-written name and bio and a few posts about the trade costs little and opens the second door.

The same goes for the other assistants born inside a social network, such as Meta AI: the platform is the source they have to hand, the website the one you control. The overall picture is in the glossary entry on GEO.

The Grok rules doing the rounds, and what the sources say

Several rules taken as fact about Grok’s answers do not appear in the X and xAI documents read on 10 November 2026. The table sets each claim beside what the primary sources say.

What you readWhat X and xAI saySource
Grok reads X first, then the web

Grok decides whether to search X and the web; no order is declared.

X Help

The crawler is called GrokBot

No crawler name published by xAI.

docs.x.ai

Grok uses Google or Bing

The web search engine is not declared.

docs.x.ai

Viral posts count for more

No rule on the weight of posts or engagement.

X Help, docs.x.ai

A Grokipedia article gets you in

No document links Grokipedia to Grok’s answers.

grokipedia.com, docs.x.ai

The web mode is called DeepSearch

The name does not appear in the xAI pages read today.

docs.x.ai

What I measured on itria.io

I have no measurements on Grok. I have them on the assistants that document their doors, and the faults I found concern the website, so they hold for any reader.

The email that wasn’t there. Cloudflare’s obfuscation stripped the address from the HTML and put it back with a script: to an assistant, the contact page had no contact details. The same fault on 14 producers’ sites out of 121, in the piece on the email AI assistants can’t see.

From one page to five. On 25 September 2026, for Claude’s web search, itria.io had a single page in the morning and five in the evening, after I asked Brave to recrawl seven: in Brave and Claude.

The result. Between 11 October and 7 November 2026 Bing’s AI Performance report counted 268 citations of itria.io pages in Copilot, and Search Console 3,550 impressions in AI Overviews and AI Mode.

How can I tell if Grok cites me?

Put to Grok the question a customer would ask, on X and on grok.com, and note which sources it cites and on what date. xAI offers no dashboard for publishers, unlike Bing Webmaster Tools for Copilot and Search Console for Google, so the manual test is the only direct measure.

Write the question the way a customer would: “who repairs boilers in Bari”, “organic olive oil producers in Puglia who export”. Repeat it every month with the same words, and check whether the company’s posts or its website appear among the sources. The full method, with the tools for the other assistants, is in the guide on how to measure GEO.

What counts stays the same as elsewhere: open doors, readable copy, facts with their source, a name written the same way everywhere. How to build this for a real business, with the questions worth owning, is told on the page about presence in search engines and AI assistants. The other guides in the series are in the SME marketing index, and the vocabulary in the entry on AI crawlers.

Questions and answers

How do I show up in Grok?

By keeping two doors open. On X, a public business account whose name, bio and posts state your trade and your town: Grok searches public posts and profiles. On the web, pages that open without blocking unknown clients, with the facts and their sources written into the text.

There is no form for signing up to Grok, and xAI publishes no rules on how it picks its sources.

Does Grok read my website?

It can. The xAI documentation describes a tool Grok uses to search the web in real time and open pages, and it returns the sources along with the answer. xAI does not say which search engine it relies on, though, or what its crawler is called: for your site the general rules apply, a page Grok can read and an engine can index.

How do I block Grok in robots.txt?

xAI does not document it. In the xAI documentation read on 10 November 2026 the words robots.txt and crawler never appear, and neither does a bot name. The rules for GrokBot or xAI-SearchBot doing the rounds come from third-party lists, which disagree with each other on the names.

Do I need an X account to show up in Grok?

To show up with your own posts, yes, and the account has to be public: X says protected posts are not surfaced in Grok’s answers. Without an account you still have the web, meaning your site and other people’s pages about your business.

Does Grokipedia matter for showing up in Grok?

No xAI document says so. Grokipedia is an encyclopaedia written by Grok, launched on 27 October 2025, where readers propose edits that Grok checks against sources and approves, and anyone with an xAI account can suggest a new article. Which companies deserve an article is not written down anywhere.

Notes on sources

  1. X Help Center, About Grok, no date on the page; read in the archived copy of 1 November 2026, because the live page does not answer automated clients. Source of the two searches, protected posts, the training setting, availability by country and the first quotation.
  2. xAI, developer documentation: Web Search (allowed and excluded domains, five at most; sources returned) and X Search (keywords, semantic search, users, threads). Read on 10 November 2026.
  3. xAI, full documentation in a single file, searched on 10 November 2026 for “robots.txt”, “crawler”, “user-agent” and the bot names in circulation: no results. x.ai robots.txt, read the same day; the Content-Signal syntax is described on contentsignals.org.
  4. Grokipedia, home page and Suggest page, read on 10 November 2026: version 0.3, count of approved edits, declared workings.
  5. Harold Triedman, Alexios Mantzarlis, What did Elon change? A comprehensive analysis of Grokipedia, arXiv:2511.09685, 12 November 2025: launch date and findings from the abstract.
  6. Data Protection Commission, inquiry into Grok training, 11 April 2025, source of the second quotation; inquiry into images generated with Grok, 17 February 2026. No decision found as of 10 November 2026.
  7. For comparison: OpenAI, Overview of OpenAI Crawlers; Perplexity, Perplexity Crawlers; Anthropic, crawler page.
  8. The itria.io figures come from Search Console (generative AI report) and Bing Webmaster Tools, window 11 October to 7 November 2026.
·Who does it for you

Opening your profile and your site to Grok takes twenty minutes. Getting named ahead of your competitors is the work after that. Positioning is built starting from the market.

Mattia Esposito, who wrote this guide, also works with Itria on positioning: in-depth market analysis to find where competitors are exposed and take that ground. He did it on his own site first. In August, asked who had founded itria.io, ChatGPT couldn’t find him. From 11 October to 7 November 2026 itria.io pages appeared 13,803 times in Google and Bing results and 3,818 times in AI answers: Google’s AI Overviews and AI Mode, and Microsoft Copilot.