What Does ChatGPT Say About Your Brand? How to Check

Here's a question most founders ask once and then forget: what does ChatGPT actually say about your brand when someone asks?

Not what you've told ChatGPT. Not your official description. What it says when a potential customer types "what's the best [tool] for [problem]" and hits enter.

We've run this check for hundreds of brands. The results are often surprising — and not in a good way. Companies with strong reputations get described neutrally, or not mentioned at all. Competitors with half the product quality show up first. Brands spend years building SEO presence and have zero AI presence to show for it.

This article walks you through how to check manually, what the results tell you, and why the check needs to be more than a one-time curiosity.

How to check what ChatGPT says about your brand (step by step)

This takes about 20 minutes and you don't need any tools. Just a ChatGPT account.

1. Write down 10 questions your customers are actually asking.

Don't start with your brand name. Start with the problem your customers solve. A project management SaaS would write: "What's the best project management tool for remote teams?" A bookkeeping service might write: "What accounting software do consultants recommend?"

These are the questions your buyers are asking AI right now. Your goal is to see whether you appear when they do.

2. Open a fresh ChatGPT session and run each query.

Use GPT-4o (the default model). Paste the first question and read the full response. Does your brand appear? If yes — where? First mention, second mention, buried in a list of six?

Copy and paste the response into a notes document. You'll want to refer back later.

3. Run the same query again, twice more.

ChatGPT responses are non-deterministic. The same question can produce different results across sessions because of how the model samples answers. If your brand appears in 1 out of 3 runs of the same query, that's meaningfully different from appearing in all 3 — or none.

4. Vary the query phrasing.

"Best tools for remote project management" and "project management software for distributed teams" sound similar to a human but may produce different AI responses. Test both ways your buyers might phrase the question.

5. Note how your brand is described, not just whether it appears.

When ChatGPT mentions your brand, what does it say? Does it get your positioning right? Is the framing positive, neutral, or hedged? "X is popular but has a learning curve" is a very different mention from "X is well-regarded for its ease of use."

6. Run the same queries on Gemini and Perplexity.

Results vary across models. A brand that's invisible in ChatGPT may appear prominently in Perplexity, and vice versa. You need coverage across all three to understand your actual AI footprint.

What the results actually mean

Once you've collected your results, you'll likely fall into one of four situations.

Not mentioned at all. This is the most common outcome for smaller brands. It doesn't mean ChatGPT has a negative opinion of you — it means you don't have enough reliable signal in the data it was trained on. You're invisible to a meaningful slice of potential customers who use AI for research.

Mentioned inconsistently. You appear in some queries but not others, or in some runs but not all. This is actually common, and it's more actionable than complete absence. It means you're in the model's knowledge but not strongly anchored to the specific problem your customers are solving.

Mentioned but positioned late. You appear, but third or fourth after competitors. For buyers, AI recommendations work a lot like search results — the first mention gets disproportionate attention. Showing up at position 4 of 5 is not the same as showing up first.

Mentioned with wrong or outdated framing. Your brand appears, but the description is inaccurate, focuses on an old product direction, or leads with a weakness. This can be more damaging than not appearing, because buyers come away with a wrong impression they believe came from a neutral source.

Why manual checks aren't enough

What you just did above is genuinely useful. But it has four real limits that matter.

You can't track change over time. The check you did today tells you nothing about whether your AI visibility is improving or declining. ChatGPT gets retrained. Competitors publish new content, get press coverage, earn more community mentions. Your relative position shifts. Without recurring checks, you'll never know.

10 prompts is a tiny sample. Your customers ask about your product category in dozens of different ways. "Best CRM for small businesses" and "affordable CRM for teams under 10 people" might produce completely different results. A meaningful visibility picture requires 20, 50, 100+ prompt variations — run systematically, not once.

You're only checking one model. ChatGPT is one AI. Gemini runs on Google's model, Perplexity uses live web retrieval, Claude has its own training data. A brand that's visible in one may be absent in another. Understanding how different models treat brand information changes where you invest.

Competitors don't stand still. While you're running a manual check every few months, a competitor might be actively improving their AI visibility — publishing authoritative content, earning press mentions, building community presence in the forums that LLMs learn from. You need to track your competitors' AI mentions, not just your own.

The gap between awareness and data

Most founders who do this manual check come away with one of two reactions.

The first: "Oh, we appear, we're fine." And then they close the tab and don't check again for six months.

The second: "I had no idea this was a problem. Now what?"

If you're in the second group, the useful next step isn't more manual checks. It's a systematic baseline — run once, automatically, across the queries that matter for your category. See where you stand across models, what position you occupy, how competitors compare. That baseline is what makes the subsequent weeks' data meaningful.

The issue isn't that manual checks are wrong. It's that they're too infrequent and too narrow to catch the changes that matter. A competitor could gain 20 percentage points of AI mention share in two months. You'd never know from a quarterly spot check.

What good AI brand visibility looks like

For reference, here's what strong AI brand presence actually looks like in practice:

You appear in over 60% of relevant queries across at least 2 major models. Your brand is mentioned in the first or second position in list-format responses. The description ChatGPT provides matches your current positioning — not an old version of your messaging. Your brand appears even in the more specific, long-tail queries ("best CRM for freelance designers") not just generic category ones.

Getting there isn't mysterious. The signals that drive AI brand recommendations — training data authority, category-query alignment, community sentiment — are all things you can influence. But you can't improve what you're not measuring.

The faster way to get this data

Running the full manual process above once is worth doing. It builds intuition. But doing it every week, across 50+ prompts, across three models, while tracking competitor movements — that's a job in itself.

BrandPulse runs your full prompt list weekly across ChatGPT, Gemini, and Perplexity, tracks mention rate, position, and sentiment, and sends you a report comparing this week to last. No dashboard to check. The report lands in your inbox and shows you exactly what changed — including whether a competitor gained ground.

The free audit runs once and gives you a baseline: your mention rate across relevant queries, what AI says about you, and how you compare to the top three competitors in your category. No account needed.