How do you get Grok to recommend your brand?

The first source is training. xAI, the company that makes Grok, says Grok was pre-trained primarily on publicly available information, including raw web page data and text extracted from the internet. X says it may share public data from X with xAI to train Grok: public posts and the engagement on them, public Spaces, and public profiles such as a bio and display name. xAI publishes the knowledge cutoff of its newest model in its API documentation, and a model's training covers only what was published before its cutoff, on your site, on X and everywhere else.

The second is search. X says Grok can decide whether or not to search public posts on X and run a real-time web search when it answers, and xAI's product page says it searches the web and X live. In xAI's API documentation, web search lets Grok search the web and browse pages, and X search lets it search posts by keyword and by meaning, look up users and fetch whole threads. xAI's privacy policy says the internet search results its service uses are provided by internet search providers, which it does not name, and the public posts by X.

For a brand, that makes its own pages and its own X account the quickest levers: search can pick up a change once it finds the page or post, with no wait for a new model.

Access is where Grok differs from the other assistants. xAI's documentation of Grok's web search says it searches the web and browses pages, and names no crawler, no user agent and no robots.txt rule for doing it, so there is no Grok line to add.

What a robots.txt file can still do for Grok is indirect. The web results come from search providers xAI does not name, so no single crawler can be called Grok's, and the most your file can do is stay open to search engines in general. That advice is ours rather than xAI's. Which crawlers to allow for the other assistants, and why a rule at your network edge can refuse one your file allows, is its own answer.

Then your account on X, which Grok reads in two ways. X says Grok can search public posts when it answers, and that public posts and public profiles are data it may share with xAI for training. So what the brand posts, and what its profile says the company does, is written by the brand and can be read by Grok. Two X settings change that, and they do different things:

• Training. In X's settings, under Privacy & Safety, then Data sharing and personalization, then Grok & Third-party Collaborators, an account can deselect sharing its public data for training and fine-tuning. X says the setting applies to training and fine-tuning by xAI.
• Protected posts. X says protecting an account's posts keeps them out of Grok's training and stops them being surfaced in answers to people's questions. Protected posts are also no longer visible to everyone.

Which to choose is the business's decision. The difference that matters here is that X describes only the second as keeping posts out of answers.

Then the pages themselves, and here the advice is ours rather than xAI's. Buyers usually describe a situation rather than name a product: a small team, a budget, a named use. Answer that situation directly on the page, in a passage that still makes sense read on its own.

And the facts in both places. xAI's consumer FAQs say that because Grok was trained on publicly available information, which can include misleading or inaccurate information, its answers can repeat it. Your own pages and profile are among what Grok can read, so a wrong price, date or specification there can reach an answer. Where an answer and your page disagree, the fix can sit on either side. Which of the three to start with depends on where your brand is weakest, and that order has its own answer.

You can see which of those sources Grok uses in your own category. Ask a buying question and open the sources it cites: xAI's product page says Grok gives live citations from primary sources across the web. Where the same sites or accounts keep coming back on questions a competitor wins and you do not, that is where the work goes. Why a competitor comes up first has its own answer.

Check a source before acting on it. X's help page says Grok may confidently give incorrect information, summarise wrongly or miss context, and asks readers to verify what it says.

Those sources sit at different distances from you, and the work changes with the distance:

• A listing you supply. A retailer page, a marketplace listing, a directory profile. You can influence it directly by correcting what you supply, though the page is not yours.
• An independent review or comparison. You can give the writer accurate facts, access to people who can answer questions and, where it fits, the product to try. What they conclude is theirs.
• Someone else's post on X, or a forum thread. You can answer the question openly, under the brand's own name. What people say about you is theirs.

The engagement on those posts is part of what xAI uses. Its privacy policy says its service uses public posts on X together with engagement data such as followers, views, likes, reposts, shares and replies, and it does not say how any of it is weighed.

None of it moves on your schedule. A page or post about you changes when the people who publish it decide to change it.

X's help page on how X ads work describes ads on X, including Promoted Ads and Spotlight Takeover, each clearly marked with an Ad label, and does not mention Grok. Its list of advertising formats, from Promoted Ads to takeovers, has no placement inside Grok, and xAI's pages describe no advertising in Grok at all.

So an X ad buys a labelled slot in front of people on X. A brand Grok's answer leaves out is still left out, and a report that counts ads and recommendations together overstates how often the brand is recommended.

For ChatGPT, OpenAI publishes where its ads appear and says advertisers cannot shape its answers, which the ChatGPT answer sets out. Neither xAI nor X publishes anything of the kind for Grok, because neither describes ads in it.

The variation is large. In Delphi's study of answers collected up to 24 August 2026, an assistant asked the same question more than once inside one measurement named the identical set of companies 21.5% of the time. A before and after built from one answer each compares two draws, and it can show a gain or a loss that nothing caused. How much evidence a change needs has its own answer.

The free check by hand, written for ChatGPT, sets out which questions to ask and what to write down for each answer, and the same check works in Grok. Comparing a before with an after adds to what you record each time. Three things are specific to doing it in Grok:

• Keep your own history out. X says Grok on X can be personalised from your X data, such as your posts, your interests and how you use Grok, and xAI says the same of its own app and grok.com if you allow it. Switch it off before the first question: on X, the Grok Personalization option under Grok & Third-party Collaborators; in the app, “Personalize Grok using X” under Settings, Data Controls; on grok.com, the same option under Settings, Data.
• Ask in the same place. Grok on X and Grok in its app or at grok.com each have their own settings, and xAI's privacy policy says it may work out the general area a device connects from, using information such as its IP address. Ask the before and the after in the same one, from the same place and network.
• Allow for the clock. There is no robots.txt change to wait for. A rewritten page can only show up once search finds it again, a third-party page once its publisher changes it, and a change in what the model learned only once a model with a later knowledge cutoff replaces it. X describes Grok's access to public posts as real time.

Traffic shows only part of it: an answer that recommends you and sends nobody leaves no visit at all.

The study publishes that figure with the counts behind it and how each was made. A brand's standing in Grok is one part of its AI visibility. Some of the work itself carries across and some does not. Accurate pages and third-party coverage are there for any assistant that reads the web. Crawler access is set operator by operator, and for Grok there is no crawler to set.

The same question has its own answer for ChatGPT, Gemini, Claude and Perplexity.

Delphi measures them side by side. Each time it measures a brand, it puts the same questions to ChatGPT, Gemini, Claude, Perplexity and Grok more than once and keeps every answer word for word. Its score, the Delphi Index, carries a margin set by how far the assistants disagreed with each other on your questions, and the dashboard and the history page do not report a change in the Index smaller than that margin as a change. Each assistant's share of answers naming you is shown beside the others' from one measurement to the next, so a change in Grok alone is visible, though that figure carries no margin of its own.

Delphi asks each assistant through its API, the route developers use, rather than through the consumer app. xAI says Grok has no knowledge of current events beyond its training data unless its search tools are switched on, and most of Delphi's questions reach Grok with them off. So a change to your pages shows in a Delphi reading first where an assistant searches as it answers, and in most of its Grok answers only once a newer model has learned it. Where Delphi does search a Grok question, it switches on Grok's web search and not its X search, so its readings of Grok do not include what an X search would find. Its check of your robots.txt has no line for Grok, for the reason above. Each paid level includes one or two measurements per brand a month. It is not a daily tracker, and an extra measurement costs $49 when a before and after needs one sooner.