How to Track Brand Mentions on ChatGPT

If you ask ChatGPT "What's the best CRM for a small accounting firm?" five times, you can get five different answers. Your brand might be named first in one, third in another and missing from the rest. So "Does ChatGPT mention us?" doesn't have a yes or no answer. It has a rate.

This guide shows you how to track brand mentions on ChatGPT the way you'd track any other channel: a fixed set of prompts, repeated samples, a log you can trend over time, and the traffic it sends in GA4. You can do all of it by hand with a spreadsheet. We'll also show where a ChatGPT rank tracker saves the work.

Why tracking ChatGPT brand mentions is different from Google rankings

On Google, a keyword has a results page and your site has a position on it. Rank trackers have worked that way for 20 years. ChatGPT doesn't work like that:

• The answer changes from run to run. ChatGPT generates each answer fresh, so the brands it names, and the order it names them in, vary even when the prompt is identical.
• There's no list of ten blue links. Brands are named inside sentences, bullet points or tables. "Position" has to mean where your brand appears in the answer, relative to the other brands named.
• It depends on who's asking. Memory, custom instructions and location can all shape the answer. Your own ChatGPT account is a biased sample.
• It doesn't always search. ChatGPT decides for itself whether to search the web for a prompt. When it does, the pages it reads and cites have a big influence on which brands it names.

Here's what that looks like in practice. These are ten runs of the same prompt, with the brand's position in each answer:

1. Run 1#2
2. Run 2Not named
3. Run 3#1
4. Run 4#3
5. Run 5#2
6. Run 6Not named
7. Run 7#2
8. Run 8#4
9. Run 9Not named
10. Run 10#2

A screenshot of run 3 would make the client happy. A screenshot of run 6 would start a fire drill. Neither is the truth. That's why every number you report should come with its n: how many answers it's based on.

What to track: mentions, positions, citations and the searches behind them

Before you run a single prompt, decide what you'll record for each answer. These are the metrics that matter, and they work the same way whether you track by hand or with a tool:

A mention and a citation aren't the same thing. ChatGPT can recommend your brand without linking to your site, and it can cite your blog post while recommending a competitor. Track them separately.

How to track brand mentions in ChatGPT manually, step by step

You don't need a tool to start. A spreadsheet and an hour a week will give you a baseline. Here's the process.

Step 1: Build a prompt list from real customer questions

Prompts are to ChatGPT what keywords are to Google, but they're longer and more conversational. Aim for 10 to 20 to start, across three types:

• Category prompts: "What's the best project management tool for a small agency?"
• Problem prompts: "How do I stop clients missing invoice deadlines?" These are how people describe your industry in their own words, before they know which product they need.
• Comparison prompts: "Asana vs Monday for a 10-person team." Include your competitors by name.

Good sources: sales call notes, support tickets, your Google Search Console queries phrased as questions, and "People also ask" boxes. Don't include your own brand name in category prompts. You want to know if ChatGPT recommends you unprompted.

If you're tracking whether ChatGPT recommends your SaaS product, write the prompts the way a buyer would at the shortlist stage: "Which [category] tools integrate with [tool you integrate with]?" or "What's a cheaper alternative to [market leader]?"

Step 2: Run each prompt in a clean session

Your own ChatGPT history skews the results. Use a Temporary Chat and choose the unpersonalized option if you're offered one, or use a logged-out browser window. Either way, ChatGPT won't draw on your memories or custom instructions. Start a new chat for every run, so earlier answers don't influence later ones.

Step 3: Run each prompt more than once

This is the step most people skip, and it's the one that makes the data worth anything. Run each prompt at least 3 times per check. More is better: 3 runs can tell you whether you're usually named, but not whether a change from 2/3 to 1/3 is real.

Step 4: Log every answer

Use one row per run. Here's a simple log you can copy into a spreadsheet:

Count position among the brands named in the answer: if ChatGPT names Ledgerly, then you, then Bookwise, you're #2. Save the full answer text and the cited links too. You'll want them later when you work out why a competitor keeps getting named ahead of you.

Step 5: Turn the log into rates

At the end of each week, roll the log up per prompt: mentioned in X of N runs, your typical position, and the competitors named most often. Report "seen in 7 of 10 answers, usually #2", never just "yes, we're in ChatGPT".

Where manual tracking breaks down: 20 prompts × 3 runs is 60 chats a week, each one read and logged by hand. Add a second market or a few clients and it stops being practical. Small samples also make week-to-week changes hard to trust.

How to track ChatGPT rankings over time

Tracking brand performance in ChatGPT responses over time is where automation pays off. A ChatGPT rank tracker runs the same prompts on a schedule, samples each one several times, and keeps every answer, so you get trends instead of snapshots.

This is what SERPent's ChatGPT rank tracker does. It sends your prompts to ChatGPT as written, multiple times every day, and reports:

• Your position among the tracked brands in each answer, with the change since the previous period.
• Your seen rate as answers out of n, so you always know how many answers a number is based on.
• Every competitor named in the same answers.
• The pages ChatGPT cited, and the searches it ran to build each answer.

ChatGPT United States · last 7 days

Prompt What's the best CRM for a small accounting firm?

Brand	Position	Change	Seen
Ledgerly	#1	–	90%19/21
ClientNestYou	#2	▲ 1	71%15/21
Bookwise	#3	▼ 1	62%13/21
Tallyhub	#4	–	29%6/21
Practix	#5	new	10%2/21

When you read the trend, look at weeks, not days. A brand going from 15/21 to 13/21 seen is within normal variation. A brand going from 15/21 to 4/21 and staying there is a real change, and the cited pages and searches usually show you why.

ChatGPT is reported on its own, not averaged with Google, because the brands it recommends can be very different. If you also want to know how your brand shows up in Google AI Overviews or Google AI Mode, track those as separate channels with the same prompt set.

Comparing tools? We've put together sourced comparisons of PromptWatch alternatives, Peec AI alternatives, Mentions alternatives and Profound alternatives, including tools that cover Perplexity, Claude and Gemini as well, plus a head-to-head of Profound vs Peec AI.

How to track which pages ChatGPT cites

ChatGPT citations are the links it shows with an answer when it has searched the web. Tracking them tells you two things: whether your own pages (blog posts, product pages, comparison pages) are being used as sources, and which third-party pages are shaping the answer.

To track ChatGPT citations by hand, open the sources for each answer in your log and record every URL. Over a few weeks, patterns appear:

• Your own pages that get cited. These are working. Keep them fresh and accurate.
• Listicles and reviews that cite competitors but not you. These are your best outreach targets.
• Directories, marketplaces and forum threads. ChatGPT often leans on these for recommendations, so getting listed or joining the conversation can move the needle.

How to track client brand visibility in ChatGPT (for agencies)

If you're tracking ChatGPT brand mentions for clients, the method is the same, with three additions:

1. Agree the prompt list with the client. It's easier to report against prompts they've signed off than to defend your choices later.
2. Track their competitors by name. "You're seen in 60% of answers" means more next to "your top competitor is seen in 85%".
3. Show the n in every report. Clients will screenshot ChatGPT themselves and get a different answer. Showing that your numbers come from dozens of answers, not one, heads off that conversation.

SERPent was built by an SEO agency for exactly this, with each client's ChatGPT, AI Overviews and AI Mode visibility kept separate. See how it works for agencies.

How to track ChatGPT traffic in GA4

Mentions are only half the story. If ChatGPT cites your pages, some readers click through, and you can see those visits in Google Analytics 4.

Start with the built-in channel. GA4's default channel group now includes an AI Assistant channel for visits from sources like ChatGPT, Gemini, Copilot and DeepSeek. Open Reports → Acquisition → Traffic acquisition and look for the AI Assistant row. Google's own AI Overviews and AI Mode aren't in it; they stay in Organic Search (see how to track AI Mode traffic).

To see ChatGPT on its own, set up a custom channel group, since the built-in channel combines every AI assistant:

1. In GA4, go to Admin → Data display → Channel groups and click Create new channel group. You'll need Editor access to the property.
2. Add a new channel called ChatGPT.
3. Set the condition to Source matches regex chatgpt\.com|chat\.openai\.com.
4. Don't add a Medium condition. ChatGPT adds utm_source=chatgpt.com to many of the links it cites (we see it on the citation URLs we collect), and some of those visits arrive without a referral medium.
5. Move the new channel to the top, above AI Assistant, Referral and Unassigned. GA4 puts traffic in the first channel whose definition it matches, so order matters.
6. Save the group, then use it in the Traffic acquisition report. Custom channel groups apply to past data too.

To track conversions from ChatGPT, use the same channel group with your key events (sign-ups, demo requests, purchases) in the Traffic acquisition report or an Exploration. That's as close as you'll get to conversational conversion metrics today: GA4 can't see what was said in the chat, only what visitors did after they arrived.

Expect undercounting. Visits from the ChatGPT apps, or from a copied link, often arrive with no referrer and land in Direct. GA4 shows you the floor of ChatGPT traffic, not the total, so pair it with mention tracking rather than relying on either alone.

What to do when ChatGPT doesn't mention your brand

Tracking only matters if it leads to action. When you're missing from answers, the cited pages and the searches ChatGPT ran are your roadmap:

• Rank for the searches it runs. If ChatGPT searched "best CRM for accountants 2026", that's a query worth targeting on the web.
• Get onto the pages it cites. Pitch the listicles, get listed in the directories and take part in the threads it keeps citing.
• Publish the page it's missing. If competitors are cited for a comparison or a use case you don't cover, write the stronger page.
• Re-measure. Keep the same prompt set running so you can see whether the work moved your seen rate, with enough answers to trust it.

This is generative engine optimization (GEO) in practice, and it only works if you can measure ChatGPT performance consistently before and after.

How to track brand mentions on ChatGPT: FAQ