ChatGPT Brand Mentions
More buyers now open ChatGPT or Perplexity before they open Google. When they ask "what's the best tool for X" or "who should I hire for Y," the answer names a handful of brands in plain prose. If yours is one of them, you get considered. If it isn't, you never enter the shortlist, and you never see the lost opportunity in your analytics because there was no click to track.
That is the problem with AI visibility: it happens inside someone else's answer, where your usual measurement tools cannot reach. This guide walks through a practical, honest way to measure whether AI engines name your brand, what "share of voice" means in this context, and how to turn the measurement into action. No special software is required to start.
Why mentions matter more than links now
The first thing to understand is that AI answers favor mentions over links. BrightEdge analyzed how brands surface inside ChatGPT and found that the tool mentions brands roughly 3.2 times more often than it cites them, an average of about 2.4 brand mentions per prompt against 0.74 citations. In other words, for every link ChatGPT shows, it makes around three unlinked brand recommendations.
This changes what you should measure. Classic SEO trains you to count links and ranking positions. In AI answers, the unit that matters is whether your name appears in the text at all, linked or not. A buyer reading "tools worth considering include A, B, and C" forms a shortlist from those names regardless of whether any of them is hyperlinked. So your measurement has to count mentions, not just citations or referral clicks. If you want the deeper background on why engines surface names this way, see how ChatGPT picks brands.
Step 1: Define your category prompts
You cannot measure visibility in the abstract. You measure it against a fixed set of questions, the same ones your buyers actually type. These are usually called category prompts, and building a good set is the most important step because everything downstream depends on it.
Start with 20 to 40 prompts and group them by buyer stage:
- Awareness prompts ask about the problem, not the product. "How do I reduce churn in a subscription business" or "what causes slow page loads." These reveal whether you show up before a buyer even knows your category exists.
- Comparison prompts name the category directly. "Best [category] tools for mid-market teams" or "top alternatives to [well-known competitor]." This is where shortlists form.
- Transactional prompts sit closest to a decision. "Which [category] service has the fastest onboarding" or "who should I hire for [task]."
To find these, talk to your sales team about the questions prospects ask, scan communities like Reddit for how people phrase category doubts, and ask ChatGPT itself what concerns buyers raise about your category. Write each prompt the way a real person would, not the way a marketer would. Then freeze the list. Even small wording changes shift AI responses, so a fixed prompt set is what lets you compare month to month and trust the difference.
Step 2: Check the engines by hand
Before paying for anything, do this manually. It is tedious, but it teaches you exactly what your buyers see and grounds every later number in reality.
Open ChatGPT, Perplexity, and Gemini. Microsoft Copilot and Claude are worth adding if your audience uses them. For each prompt in your set, run it and record three things: whether your brand is mentioned at all, where it falls in the answer (named first reads very differently from a passing reference at the end), and which competitors appear alongside you.
Run each prompt three to five times per engine, ideally in fresh sessions. AI answers are not deterministic, so a single run can mislead. If you appear in one of five runs, your real visibility is roughly twenty percent for that prompt, not a clean yes or no. Use a plain spreadsheet: one row per prompt, columns for each engine, and a simple mention count. Repeat the whole exercise monthly so you are comparing like with like.
Step 3: Understand share of voice in AI answers
Once you have raw mention data, share of voice gives it shape. In AI answers, share of voice is the percentage of relevant prompts where your brand appears, measured against the competitors who appear on those same prompts. If twenty prompts are relevant to your category and you are named in eight of them, your mention rate is forty percent. If competitor A appears in fourteen, they are winning the category conversation regardless of who ranks higher in classic search.
Two cautions keep this honest. First, report per-engine numbers, never a blended average. ChatGPT, Perplexity, and Gemini disagree on brands a large share of the time, so a single combined figure hides where you are strong and where you are invisible. Second, weight by prompt importance. Owning every low-intent awareness prompt while losing the comparison prompts that decide purchases is not a win, even if your overall percentage looks healthy.
Step 4: The monitoring tool category
Manual checks work for a small prompt set, but they do not scale to hundreds of prompts across five engines run weekly. That is the gap a category of AI visibility monitoring tools fills. Broadly, these tools automate what you did by hand: they run your prompt set across multiple engines on a schedule, log whether your brand and competitors are mentioned, track position and citations over time, and present share of voice as a trend rather than a one-off snapshot. Many also attempt to surface which sources the engines drew on.
The category includes options aimed at enterprises and lighter tools for smaller teams, at a wide range of prices. Rather than endorse specific products, the practical point is to treat them as measurement instruments. A good one gives you a reliable, repeatable read on the inputs. For a fuller breakdown of where measurement ends and influence begins, see our look at ChatGPT brand mentions versus AI visibility tools.
Step 5: Turn measurement into action
This is the step most teams skip, and it is the only one that changes the number. A dashboard tells you that you appear in eight of twenty prompts. It does not tell you how to reach twelve. To move the result, you have to move the inputs the engines rely on.
AI engines tend to name brands that are described consistently across the wider web, in the kind of independent, third-party content that reads as a recommendation rather than an advertisement. So the work is twofold. Strengthen your own entity footprint: clear, factual pages about what you do, comparison and alternative pages, and structured information engines can parse, much of which overlaps with good SEO practice. Then build presence in the third-party content the engines read, where your brand is mentioned naturally and in context. That second lever is exactly what a ChatGPT brand mention program is built to do: place natural, unlinked references to your brand inside existing content across a publisher network, so the sources these models trust start naming you.
Measure first, then act, then re-measure against the same frozen prompt set. That loop, not any single tool, is what turns AI visibility from a guess into something you can manage.
Want to see where you stand before you start tracking by hand? Run your category prompts through ChatGPT yourself, or look at how placements work in the packages.