Every answer, in full
Six buyer questions written for your category and area, asked to three models. You see the complete list of businesses each one named and where you came in.
Your customers have started asking ChatGPT and Gemini for recommendations the way they used to ask Google. Those answers name a handful of businesses. Type your website and see whether you are one of them.
About 30 seconds. No email, no card.
We built this to answer a question about our own clients, so we ran it across the businesses we work with before we showed it to anyone. Five of the eight were never named, once, in any answer. Two more came up only occasionally. Every one of them had competitors named repeatedly in the same answers, often the same three or four names over and over.
None of them had a broken website. Most had clean structured data and nothing blocking AI crawlers. The businesses getting named were simply the ones other websites had written about.
What the check gives you
Six buyer questions written for your category and area, asked to three models. You see the complete list of businesses each one named and where you came in.
The businesses the assistants reach for in your market, ranked by how often they came up. This is usually the part that stings.
Whether your robots.txt blocks the ten crawlers that feed these assistants, and whether your pages state the facts one would need to recommend you.
A written read on where the gap actually is, with the moves that would close it and what each one is worth. Plus a short list you can start on this week.
Why this happens
Google reads your site and ranks your pages. An assistant answers from what it absorbed about your market: press, association and chamber pages, supplier and partner sites, review platforms, other people's writing. Your own site is a small part of that, which is why a good website and zero mentions is a common combination.
We have run this on businesses sitting at the top of page one for their main search who still never come up in an answer. Different system, different scoreboard, and it is possible to win one and lose the other without noticing.
Plenty of robots.txt files block GPTBot, ClaudeBot, PerplexityBot or Google-Extended because a template or a plugin put it there. Nobody made that call on purpose. The check tells you either way, so you can decide.
Getting written about somewhere other than your own site, review volume that keeps up with your competitors, and one consistent statement of who you are and where you work. Slow, unglamorous, and it compounds.
If you want help with it
Nobody can promise a model will start naming your business, and you should not believe anyone who does. What we can do is the work that earns it, which is the same work that makes you easier to find everywhere else. It runs inside an SEO engagement rather than as a separate product, because it is the same job.
Local press, trade publications, association and chamber pages, supplier and partner sites, the directories that carry weight in your category. This is the slow part and it is the part that moves the answer.
We set up the asking so it happens on its own after every job, on the platforms your category actually gets judged on. Recency counts as much as the total.
One statement of what you do, where you work, and who you serve, matching across your site, your listings, your profiles, and in the machine-readable form a model can pick up without guessing. These systems are assembling an identity for your business whether you help or not.
Crawler access decided on purpose instead of inherited from a template, and pages that answer what buyers actually ask in the words they use. Both help you in regular search at the same time.
Close, with one difference worth knowing. We ask the models from what they already know, with no live web browsing. That measures whether your business is part of what they have learned about your market, which is the part that reviews, citations, and mentions on other websites move. A live session that searches the web mid-answer can return something different. We show you the exact questions we used so you can run them yourself and compare.
Nothing. No email, no sign-up, no card. Type your website and the report loads in about 30 seconds.
It is normal, especially for a smaller or younger company. These models learn which businesses exist from what other websites say, not from your homepage. Being absent usually means there is not much written about you anywhere else yet, which is a fixable problem rather than a website problem.
It can. Plenty of sites block GPTBot, ClaudeBot, PerplexityBot, or Google-Extended without anyone deciding to, usually inherited from a template or a plugin. The check reads your robots.txt and tells you which of the ten crawlers we watch are shut out, so you can decide on purpose instead of by accident.
The work overlaps but the scoreboard is different. Google ranks pages. An assistant names businesses, and it learns which ones exist from press, directories, association pages, review platforms, and other people's writing. You can rank on page one and still never get named, and we see that regularly.
GPT-5.4 from OpenAI, Gemini 3.5 Flash from Google, and Claude Sonnet 5 from Anthropic. Six questions to each one, eighteen answers per report. The report names the models and shows every answer, so nothing about the method is hidden.