Free tool
Grok has two things to draw on that no other engine has in the same combination: an ordinary web search, and X. That second half is the reason people started asking about Grok visibility at all, and it has produced a lot of advice about posting on X to get cited.
We measured it in August 2026. Thirty-six questions, split evenly between buying questions, fact-check questions and questions about what people are saying right now. Grok cited x.com 24 times across the whole run, and every one of those 24 came from the third group. Ask it which CRM a small real estate team should buy and it reads HousingWire, Follow Up Boss and a Zapier roundup, the same sort of pages ChatGPT and Perplexity read. Ask it what designers are saying about the job market and roughly half its sources are individual X posts.
The practical version: X presence moves Grok on questions about sentiment and what is happening this week. Buyer-intent questions, the ones that end in somebody choosing a vendor, still come down to which pages are on the web about your category. If you only have budget for one of those, the pages are the one that pays.
Not mentioned, not cited. You are not in the pages Grok pulled. Look at the sources it used and pick the ones you could plausibly get into.
Mentioned, not cited. Somebody else's page named you and Grok repeated it. That is real visibility, and it belongs to whoever owns that page. It can vanish when they next edit.
Mentioned and cited. Grok read your site and used it. Note which page it took, because that page is doing the work and is worth extending.
When one of the sources is an x.com link, read the post. It tells you who is being listened to in your category, which is a useful thing to know whether or not you ever post there yourself.
A single run is one question, one engine, one day. Grok re-searches every time, and in our run the answers to identical questions moved between days. Watching whether a page you published changed anything means asking the same questions on a schedule and watching the source list rather than the yes or no. That tracking is what CueScout sells, and to be straight about the boundary: the paid product reads ChatGPT and Perplexity. Grok lives here on the free tool, where you can run it yourself whenever you want to check.
Next
The same question put to ChatGPT, which retrieves less and leans harder on training data.
The heaviest citer of the three, and the one most likely to reach for Reddit.
Which sites supply the answers in your category, ranked by share.
It sends your question to a live search-grounded Grok model, reads the answer, and reports whether your brand name appears in the text, whether your domain appears in the citation list, and which other brands got named alongside you. One question, one run, no signup. It reads the citation list the model actually returned, not URLs pulled out of the prose, because a URL written into the answer text is often something the model half-remembers rather than a page it read.
Sometimes, and much less often than the optimisation advice suggests. We put 36 questions to Grok in August 2026 and it cited x.com 24 times. All 24 came from the 12 questions that asked what people are saying right now. Across the 12 buying questions and the 12 fact-check questions it cited X zero times, and sourced the same listicles, vendor pages and news articles that every other engine reads. So if you are checking whether Grok recommends you to a buyer, you are checking its web corpus. Your X presence is a different surface and it moves a different kind of question.
No, and the difference matters. This runs a Grok model with live search on. The reply bot people summon in threads runs inside X, sees the post it was replying to, and is tuned for that context, so its mix of sources is its own. What the two share is the model and the underlying search. Treat a result here as evidence about how Grok reads your category, not as a transcript of what the bot would say in someone’s replies.
Because they read different pages. In the same 36-question run, Grok and Perplexity shared 74 of the domains they cited, and Grok and ChatGPT shared 8. Six domains were cited by all three. A page that made you visible in one engine very often does nothing in the next one, which is why a single check is worth taking as one engine’s opinion rather than a verdict.
Read the sources it did cite, which the result lists for you. Those pages are the answer you are competing with. If they are best-of lists and comparison articles that leave you out, the work is getting onto those pages, which usually means pitching the person who maintains them. If they are competitor blog posts, the work is publishing the better version on your own domain and getting it read. Most of it happens off your own site, which is the part people find irritating and is also the reason it works.
That is how these engines behave. Grok re-runs a live search each time and the results underneath it move, sometimes within a day. One run tells you whether you are plausibly in the picture. Knowing whether a change you made actually worked takes the same questions asked repeatedly over weeks, which is the paid product rather than this page.
No. The check runs and returns the result, and nothing is written to a database. Scheduled tracking runs against a brand you have added to your own account.