How to See What ChatGPT Says About Your Brand (and Your Competitors)

More of your buyers’ research now happens inside an AI assistant. Instead of scanning page one of Google, they ask ChatGPT “what’s the best tool for X?” or “is [your brand] any good?” — and they act on the answer. That answer is, in effect, a recommendation you never see and can’t currently rank-check. This guide shows you how to see it.

The short answer: to find out what ChatGPT says about your brand, (1) ask it directly with the buyer-intent prompts your customers actually use — not just your brand name; (2) repeat the same prompts across ChatGPT, Claude, Gemini, and Perplexity, because each answers differently; (3) check the sources cited to learn what’s feeding the answer; and (4) repeat on a schedule, because these answers change week to week. A manual pass gives you today’s snapshot. Catching the drift over time is where a tracking tool — like Agonai, which is our product; bias disclosed — earns its place.

Guide last reviewed: July 20, 2026.

Why what ChatGPT says about your brand matters now

For most of the last two decades, “how do buyers find us?” had one honest answer: search rankings, which you could check any time. AI assistants broke that. A growing share of buyers now ask an assistant to shortlist tools for them, and the assistant returns three or four names with reasons. If your brand isn’t in that shortlist — or is in it with a wrong detail — you lose the deal before a human ever visits your site.

Three things make this urgent:

It’s invisible by default. You can open Google and see where you rank. You cannot open ChatGPT and see where you “rank” without deliberately going to look, prompt by prompt.
It’s already influencing pipeline. Buyers treat a confident AI shortlist the way they used to treat the top organic results — as a filtered starting point.
It moves. The answer changes as models update and as the sources they read change. A competitor can appear in the recommendation set in a week, and you’d never know.

How to check what ChatGPT says about your brand (the manual method)

You can do a real audit today, for free, in about thirty minutes. Open a fresh chat (log out or use a temporary chat so your history doesn’t personalize the results) and run three kinds of prompts.

1. Brand-name prompts — what does it know about you?

“What is [your brand]?”
“What does [your brand] do, and who is it for?”
“Is [your brand] any good? What are its pros and cons?”

You’re checking for accuracy: does it describe you correctly, cite the right pricing, and avoid confusing you with someone else? Wrong facts here are a bug you can often fix by publishing clearer public pages.

2. Buyer-intent prompts — do you even get mentioned?

This is the important one, and the step most brands skip. Buyers rarely start by typing your name — they describe their problem. Ask the questions they ask:

“What are the best [category] tools for [your buyer’s segment]?”
“[Your biggest competitor] alternatives”
“[Competitor A] vs [Competitor B] — which should a small team pick?”

Then read the answer as a scorecard: Are you mentioned at all? In what position? Alongside which competitors? With what framing?

3. Write down what you see. For each prompt, note whether you appeared, your position in the list, whether the description was accurate, and which competitors showed up with you. That table is your baseline.

Illustrative example (fictional companies). Say your product is Brevora. You ask ChatGPT “best competitive-intelligence tools for small teams,” and it returns Lumicast, Northcurrent, and Pelora — no Brevora. You then ask “Lumicast alternatives,” and Brevora finally appears, but described as “enterprise-focused,” which is the opposite of your positioning. Two findings, thirty seconds: you’re absent from the category prompt, and mispositioned when you do appear. Both are fixable — but only because you looked.

Don’t stop at ChatGPT — check every assistant

ChatGPT is the one people name, but your buyers also use Claude, Gemini, and Perplexity, and the four give genuinely different answers. They’re trained differently, they retrieve from different sources, and they weight them differently. It’s common to be the top recommendation in one and completely absent from another.

Run the same prompt set through all four. Perplexity and ChatGPT’s search mode are especially useful here because they show their work (more on that next). Treat each assistant as a separate “search engine” you now have to rank in.

Check the sources behind the answer

When an assistant cites links — Perplexity does this by default, and ChatGPT does in search mode — click them. The cited pages are the levers. If the answer about your category leans on a G2 category page, a Reddit thread, and two competitors’ comparison pages, then those are what shaped the recommendation, not your homepage. Knowing the sources tells you exactly where to earn a mention: your own clearly-written pages, third-party review sites, and the comparison content buyers and models both read.

Why a one-time check isn’t enough

Here’s the catch with the manual method, and it’s the same failure mode as any DIY monitoring stack (we wrote about that in Crayon alternatives for small teams): it works once, then quietly stops happening.

The answers are non-deterministic. Ask the same question twice and you can get different phrasings, a different order, even a different shortlist. One reading isn’t a measurement — it’s an anecdote.
They drift. Models get updated. Sources get published and de-ranked. The answer that put you second this month can drop you off entirely next month, silently.
A snapshot can’t show change. The whole point is to catch the moment a competitor enters the recommendation set, or the moment a model starts naming you — and change is exactly what a single check can’t see.

Doing this properly means running a stable set of prompts, across all four assistants, on a schedule, and recording the results so you can compare over time. That’s straightforward to describe and tedious to sustain by hand — which is the honest case for tooling.

Tracking it continuously (what we built)

This is the gap Agonai was built to close, so read this knowing we make it. Our AI Visibility feature runs a set of buyer-intent prompts on a schedule across ChatGPT, Claude, Gemini, and Perplexity, and tracks, over time:

whether your brand is mentioned, and in what position;
which competitors show up alongside you, and how often (share of voice);
how you’re described, and whether the framing is accurate;
and — like every Agonai insight — what happened, what it means, and what to do about it, not just a raw score.

Pricing is published ($19–$499/mo, 14-day trial, no card), because a tool that measures your visibility shouldn’t hide its own.

Watch out for: no tool controls what an assistant says — not ours, not anyone’s. AI Visibility measures and tracks the answers; changing them is downstream work (clearer pages, earning citations on the sources models read). Any product promising to “guarantee” your spot in an AI answer is overselling.

What to do once you know

If you’re absent from the category prompts: the levers are your public content and where you’re cited. Publish clear, specific pages a model can quote, and earn mentions on the third-party sources the answers pull from. (We’ll cover the mechanics — including llms.txt and generative engine optimization — in a dedicated guide soon.)
If a competitor is winning the recommendation: read the sources behind their mentions and work out what’s feeding them.
Either way, keep measuring. The single most valuable thing here isn’t one audit — it’s the trend line.

The fastest way to start is the free manual audit above: thirty minutes, one prompt set, four assistants, one baseline table. If you’d rather have that baseline maintained automatically and watched for changes, start a 14-day Agonai trial (no credit card) and point AI Visibility at your brand and competitors.

We use Agonai’s own AI Visibility to track how Agonai shows up in AI answers — this article is us describing a thing we do daily. Spotted a claim that’s out of date? Email hi@agonai.io and we’ll fix it.