When someone asks ChatGPT "what's the best tool for tracking AI visibility," asks Claude "which agency should I hire for AEO work," or types the same question into Perplexity, something comes back. Your brand either appears in that answer or it doesn't. Most businesses genuinely don't know which.
Closing that gap is what AI brand monitoring is for: running the questions your customers ask, in clean sessions, and recording whether your brand shows up, how it's described, and who gets recommended instead. You can do it for free, and this guide walks through the whole setup.
Prefer video? I recorded the same starting point as a short walkthrough:
How to get started with AI search
Why AI brand monitoring starts with ChatGPT and Claude
There are plenty of AI platforms: Perplexity, Gemini, Grok, Copilot. So why start with ChatGPT and Claude?
> ChatGPT and Claude are very good LLMs to get started with. They represent most of the LLM market being tracked today. If you're getting started with AI tracking or AI visibility, already mastering ChatGPT and Claude will give you an unfair advantage. — Sofian Bettayeb
They handle a large share of AI-assisted searches and recommendations. But the more practical reason is that what works in ChatGPT transfers. The content strategy that gets you cited there will almost certainly improve your visibility in Gemini and Perplexity too.
Starting with 2 platforms also keeps the data manageable. You can see patterns, act on them, and expand once you have a baseline worth building on. Perplexity is the natural third: it works differently enough to deserve its own section, which you'll find below.
The manual approach (and why it breaks down fast)
You can do this without a tool. Open a private window, type a prompt, note whether your brand appeared, repeat for Claude. For 1 prompt, that takes a few minutes. For 20 prompts across 2 platforms, it's a couple of hours, and next month you won't run the same prompts the same way, so you get no trend.
The private window matters. If you've been using ChatGPT for months, your history shapes its answers. What you see is not what a cold prospect sees.
Side by side, the difference looks like this:
How to pick your prompts
The data you get is only as good as the prompts you run. 2 rules cover most of it.
1. Start with the minimum amount
Not 50 prompts on day 1. Pick a handful of questions, run them, read the answers, fix what they reveal. You can always add more once the first batch stops teaching you anything.
2. Start where you're 100% confident your brand should be mentioned
If you sell project management software for remote agencies, "best project management software for remote agencies" is a prompt you should win. If you're absent from that answer, you've found the most important gap in your AI visibility, and you found it with 1 prompt.
Generic prompts like "what are the best marketing tools" fail both rules. The category is too broad for you to expect a mention, so whatever comes back tells you nothing you can act on.
For more on building a prompt strategy: How to create AEO topics and prompts.
How many prompts you actually need
The minimum. That's the honest answer.
Every prompt you add means more tracking, more results to read, more recommendations to act on. Past a certain point the report stops being a to-do list and becomes noise you scroll past. 10 prompts you actually act on beat 50 you don't.
2 things make a small set work:
Map the user journey. Cover the questions a buyer asks on the way to you: discovering the category, comparing options, solving the problem you solve. A few prompts per stage is enough.
Build them from first-hand data. Your Google Search Console queries, your sales calls, your customer emails. These are the questions real buyers already ask you. Turning them into prompts beats guessing what people might type into ChatGPT.
AEO Copilot's free plan allows 50 prompts. That's a ceiling, not a target. Start with 10, act on what they show you, and add more only when the current set stops surfacing new work.
Tracking it with AEO Copilot (free for ChatGPT and Claude)
AEO Copilot's free plan tracks ChatGPT and Claude at no cost, no credit card required. 1 brand, 50 prompts, manual runs whenever you want.
Setup takes about 10 minutes:
- Create your brand: name, website, and up to 10 competitors
- Define your topics: each topic groups a set of related prompts
- Add your prompts: the specific questions you want to test
- Run: AEO Copilot queries both platforms in fresh, isolated sessions and records the results
Each run returns a Visibility Score, a Sentiment breakdown, a Competitive Share, and a Visibility Trend as you build up runs over time.
The clean session approach matters. Every prompt runs as a new user with no history, so the results reflect what a cold prospect sees, not what ChatGPT returns to someone already deep in your industry.
Start tracking for free →
Adding Perplexity and Google AI Overviews
Perplexity behaves differently from ChatGPT and Claude. It searches the live web for every answer and cites the pages it used, right there in the response. That makes it the engine where your content changes show up first, and the one that tells you exactly which pages are winning the answers you want. To check it manually, open a private window, ask your category question, and read the citations. If a competitor's comparison page is cited and yours isn't, that's not a mystery to solve. That's a to-do list.
Google AI Overviews work from the other direction: they're built on Google's classic rankings. If a page already ranks for the query, it's a candidate to be cited in the Overview, so this is the engine where your existing SEO work carries over most directly. It's also the one with the most visible traffic effect, because an Overview sits above the classic results and answers many searches before anyone clicks. Watching which of your prompts trigger an Overview, and whether you're cited in it, tells you where AI answers are already sitting between you and your Google traffic.
To track both on a schedule, Perplexity and Google AI Overviews are included on the Startup plan at $9/month, alongside ChatGPT and Claude. Same prompts, same clean sessions, 4 engines instead of 2.
The free ways to track AI mentions, compared
The short version: Google Alerts watches the web, not the answers. Manual checks work for a spot check and break down as a system. If you want a real baseline for $0, the free plan is the only option on this list that gives you one.
Where the answers come from, and how long results take
ChatGPT and Claude don't read the same web. When a question triggers web search, ChatGPT retrieves through Bing's index plus licensed publishers. Claude searches through Brave and cites fewer sources, each tied to an exact quoted passage. Perplexity runs its own live retrieval on every answer. When no search is triggered, the answer comes straight from training data: whatever the model absorbed about your brand before its cutoff.
That split explains the timeline. Anything that reaches a search index moves fast: fix a page, get it indexed by Bing and Brave, and search-backed answers in ChatGPT and Claude can pick it up within 2 to 8 weeks. Training data moves slowly. Getting mentioned across enough third-party sources that the model treats you as a default answer takes 3 to 6 months of consistent work.
!Stacked timeline chart: AI visibility per engine over time. Perplexity reacts in days, Google AI Overviews in weeks, Claude in months, ChatGPT in a quarter.
The split also means visibility is engine by engine. A brand that ranks on Google but not on Bing or Brave shows up in 1 engine and misses the others. We ran 1,600 prompts across 4 engines and 62% of all brand mentions appeared in only 1 of them. The full breakdown, including what wins each engine, is here: 1,600 prompts: how visible are brands in AI.
What to do with the data
Running the prompts is the easy part. Reading what comes back is the work.
When your brand appears, check how it's described. A mention isn't always a good one. "Brand X is an option, though some users report difficulty with onboarding" is a mention, but every time someone sees that response, the AI is adding friction before they've even clicked through. That's a content problem you can fix.
When your brand doesn't appear, look at who does. The competitors showing up in your place are getting cited for a reason. Read those responses and figure out what they have that you don't. It might be a clearer use case page, a more direct answer to the question, or just content structured in a way models can parse and yours isn't.
Run the same prompts consistently. 1 run tells you almost nothing. 6 to 8 weeks of runs tells you whether anything you're doing is actually moving the needle.
For the full breakdown of which metrics to track: AEO metrics that actually matter.
Frequently asked questions
What is ChatGPT brand monitoring?
ChatGPT brand monitoring means running the questions your customers ask ChatGPT and recording whether your brand appears in the answers, how it's described, and which competitors get recommended instead. Done on a schedule, it gives you a visibility trend you can act on, the AI-era equivalent of rank tracking.
Why does it matter if ChatGPT mentions my brand?
More buyers get recommendations from AI before they open a search engine or visit a website. If ChatGPT and Claude consistently recommend competitors instead of you, you're losing consideration before the comparison stage even starts, and you'd never know unless you checked. The fastest way to check is a free AEO audit.
Is tracking brand mentions in ChatGPT actually free?
With AEO Copilot, yes. The free plan covers 1 brand, 50 prompts, tracking across ChatGPT and Claude, no credit card required. Most other tools offer a short trial or put ChatGPT behind a paid tier, as the tool comparison shows.
How is this different from Google Alerts?
Google Alerts monitors web content for your brand name. AI brand monitoring tracks what language models say about your brand when someone asks a question. A brand can have strong web presence and still be invisible in AI answers. They measure completely different things.
How often should I run my prompts?
Once a week is enough. AI outputs don't shift dramatically day to day, and a single run is too noisy to mean much. The pattern over 4 to 8 weeks is what you're actually looking for, and these are the metrics worth reading.
Can I track more than 2 LLMs on the free plan?
The free plan covers ChatGPT and Claude. Perplexity and Google AI Overviews come with the Startup plan at $9/month. For most brands starting out, ChatGPT and Claude are the right focus. Once you have a real baseline there, adding the other 2 engines is 1 upgrade, not a new setup.
How do I track brand mentions in Claude?
Manually: open a fresh session in claude.ai, ask your category questions, and note whether your brand appears and who gets recommended instead. Keep in mind Claude searches the web through Brave, not Google or Bing, so its answers and sources can differ from ChatGPT's even on identical prompts. On a schedule: AEO Copilot's free plan runs your prompts on Claude alongside ChatGPT, in clean sessions, and records mentions per prompt.
How do I track brand mentions in Perplexity?
Manually: ask your category questions in a private window and read the citations, since Perplexity shows its sources in every answer. On a schedule: the Startup plan runs your existing prompts across Perplexity and Google AI Overviews alongside ChatGPT and Claude, so mentions and cited pages are recorded per prompt.
What if my brand doesn't appear at all?
That's still useful. Zero visibility usually means either a content gap (the AI doesn't have enough to work with to cite you) or a technical gap (crawlers can't access your site properly). The technical audit covers the second. For the first: how to show up on ChatGPT.