ChatGPT Brand Mentions
When someone asks ChatGPT or Perplexity "what's the best tool for X," a short list of brands gets named. The rest don't exist. The gap between those two groups is not luck, and it is not your ad budget. It is a set of signals that AI models read off the open web before they ever write an answer.
This is a working checklist. Roughly a dozen items, each with why it matters and the quick action to take. Run through it once, honestly, and you'll usually find two or three soft spots that explain why you're invisible in answers your competitors win. Start there.
How AI answers actually pick brands
Two findings frame everything below. First, AI engines lean on mentions, not just links. BrightEdge research found ChatGPT names brands far more often than it cites a source URL for them — the model is reading language, context and frequency across the web to decide which names belong in an answer, not counting backlinks. Second, the content that gets pulled into answers is overwhelmingly list-shaped: a Search Engine Land analysis of around 25,000 URLs found listicles dominate AI citations by a wide margin, making up the majority of the most-cited pages.
So the job is to make sure your brand name appears, accurately and often, inside the kinds of pages models trust — and that nothing technical is hiding you. That is what the checklist covers. For more on the underlying mechanics, see how ChatGPT picks brands.
The checklist
Work top to bottom. The first four items move the needle most for most brands.
- Are you named in the category roundups and "best of" listicles? This is the single highest-leverage item. Because list content dominates AI citations, being absent from the "best [category]" articles in your space is the most common reason a brand never gets named. Quick action: run your top buying queries through ChatGPT and Perplexity, note which roundups and listicles get cited, and check whether your brand appears in them. Build a target list of the ones that don't include you.
- Do you have third-party coverage, or only your own website? Models weight independent sources heavily. If every mention of your brand traces back to your own domain, you read as a brand talking about itself, not one the market discusses. Quick action: search your brand name plus "review," "alternative" and "vs" and see how many results are on sites you don't control. If the answer is "almost none," that gap is your priority.
- Are you present on the review platforms models trust? G2, Capterra, TrustRadius and similar sites get cited because they hold structured, comparative review data — exactly what models want for buying questions. Quick action: claim your profile on the two or three platforms that matter in your category, fill out every field, and ask a handful of real customers for recent reviews.
- Is your name, address and core entity data identical everywhere? Models build an internal picture of who you are from repeated, consistent details across the web. Conflicting names, URLs, founding dates or descriptions make you a fuzzy entity the model can't confidently name. Quick action: pick one canonical company name, one-line description and set of facts, and make your site, social profiles, directories and your Wikipedia/Wikidata entry (if any) all match it.
- Is your structured data clean and accurate? Schema markup (Organization, Product, FAQ, Review) gives crawlers machine-readable facts about you instead of forcing them to guess from prose. It won't manufacture a mention, but it removes ambiguity about what you are. Quick action: add or fix Organization and Product schema, validate it, and make sure it agrees with the visible page. A quick view-source check will flag missing markup.
- Can AI crawlers actually reach your pages? A surprising number of brands are invisible because their content is blocked — sometimes by their own robots.txt, sometimes silently at the CDN. Some content delivery networks and firewalls now block known AI crawlers by default, overriding an otherwise-correct robots.txt file. Quick action: check robots.txt for disallows on GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot and Google-Extended, and confirm your CDN or firewall isn't blocking them upstream.
- Have you added an llms.txt file? An llms.txt file is a plain-text map that points AI systems at your most important, answer-ready pages. It's an emerging convention rather than a guarantee, but it's cheap and signals intent. Quick action: publish a short llms.txt at your root linking your key product, pricing and documentation pages with one-line descriptions.
- Is your content answer-shaped? Models lift passages that directly and self-containedly answer a question. Walls of narrative copy with the answer buried in paragraph six get skipped. Quick action: lead key pages with a clear question as a heading and a tight, factual answer in the first two sentences, then expand. Use lists and comparison tables where the query is comparative.
- Is the information fresh? Citation patterns shift fast, and stale pages lose ground. Independent tracking has shown listicle citations in ChatGPT swinging sharply month over month, which means today's win isn't permanent. Quick action: set a recurring review on your top answer pages — update figures, dates and product details, and add a visible "last updated" signal.
- Do you have genuine review velocity and recency? A trickle of recent, detailed reviews reads as an active, real brand. A pile of three-year-old reviews reads as a brand that may not be around. Quick action: build a light, ongoing process to request reviews after a positive customer moment, so recency never dries up.
- Is your expertise attributable to real people? Models favour content with clear authorship and credentials, especially in higher-stakes categories. Anonymous content is easier to discount. Quick action: add real author bios with credentials, link to authoritative profiles, and make sure your team is associated with the topics you want to be named for.
- Are you being talked about where your buyers gather? Perplexity in particular leans on community sources like Reddit, and discussion threads frequently feed answers. A brand that never comes up in real conversation is a brand the model rarely surfaces. Quick action: identify the two or three communities and forums where your category is debated, and make sure your brand is part of those genuine conversations.
Reading your score
If you can confidently tick the first four — roundups, third-party coverage, review platforms and consistent entity data — you're ahead of most. If those are the gaps, no amount of schema or llms.txt tuning will fix it, because the technical items only help once there is something for models to find. Fix presence first, then plumbing.
This split also explains why purely on-site SEO often stalls in AI search: it controls your own pages well, but says nothing about how you appear across everyone else's. We unpack that difference in ChatGPT brand mentions versus SEO and in the tooling comparison at ChatGPT brand mentions versus AI-visibility tools.
Where most of the work actually lives
Look back at the checklist and notice how many items live off your own website: the roundups, the third-party coverage, the conversations, the consistent way you're described across other people's pages. That's the hard part, because you can't simply publish it yourself. It has to exist out in the content models already read.
That off-site presence is precisely what a ChatGPT brand mention program is for — getting your brand named naturally and accurately inside existing, relevant content across the web, so the models keep encountering it. No links, no ad spend, just more of the contextual signal that decides which names make the cut. Different industries weight these items differently; see the playbooks for SaaS and other verticals.
Want to know which boxes you're missing right now? Work down this checklist and start with the first gap it surfaces.