How to Track Your Brand Mentions in ChatGPT, Perplexity, and Claude
The simplest way to know if an AI recommends your brand: open ChatGPT, type "best [your category] tool for [your use case]", and look for your name.
That works. But it doesn't scale, it doesn't track competitors, and it gives you a snapshot instead of a trend.
Here's a structured approach — from manual to automated.
Step 1: Define your prompts
The quality of your AI visibility tracking is directly tied to the quality of your prompts. You want prompts that mirror what your actual customers would type.
For a CRM tool, that might be:
- "best CRM for early-stage startups"
- "what CRM should I use for a small sales team"
- "CRM alternatives to Salesforce for SMBs"
For a design tool:
- "best Figma alternative for product teams"
- "UI design tools for solo designers"
- "easiest prototyping tool for non-designers"
Aim for 10–15 prompts. Mix intent types: recommendation ("best X"), comparison ("X vs Y"), and problem-first ("how do I do Z" — where your product is the answer).
Step 2: Run them across all major LLMs
Don't just check ChatGPT. Each LLM has different training data and different web search behavior. You might appear prominently in Perplexity and not at all in Gemini — and that asymmetry is worth knowing.
The four platforms to cover:
- ChatGPT (GPT-4o with browse) — highest usage, web search enabled by default
- Perplexity — search-native, cites URLs in every response
- Claude (claude.ai) — strong reasoning, web search available
- Gemini — Google's model, index access
For each platform, run your prompts in a fresh session (no conversation history that might bias the response).
Step 3: Log what you find
For each run, record:
- Mentioned? Yes/No
- Position — first, middle, or end of response
- Sentiment — positive, neutral, or qualified ("it's okay for X but not Y")
- Cited URL — which URL did the model link to when mentioning you?
- Competitors mentioned — who else appeared, and in what order?
A simple spreadsheet works for this. Log the date so you can track change over time.
Step 4: Spot the patterns
After a few rounds of manual tracking, you'll start to see patterns:
- Which prompts reliably surface you?
- Which LLMs mention you more than others?
- Which competitor consistently appears above you?
- Which pages do the models cite when they mention you?
That last point — the cited URL — tells you which content is doing the heavy lifting for your GEO. If Perplexity always cites a specific G2 page when mentioning you, that page is your most valuable GEO asset right now.
The problem with manual tracking
Manual tracking is the right starting point. But it breaks down fast:
- It's time-consuming (40–60 minutes per round)
- You'll skip it when you're busy (and that's when things shift)
- You can't track weekly cadence reliably over months
- Comparing apples-to-apples across time is hard when you're logging manually
The patterns that matter in GEO are 30–90 day trends, not single data points. A one-time snapshot tells you where you are today; the trend tells you whether you're gaining or losing ground.
Automating your AI visibility tracking
This is why we built TraccoAI. You add your brand, your competitors, and your prompts once — and we run them against all four LLMs every two days.
Every run is stored with the full LLM response, your mention position, sentiment, and every URL the model cited. The dashboard shows your appearance rate over time, your share of voice vs. competitors, and which prompts are performing.
You get a weekly email digest so you know your trend without having to log in constantly.
It's the difference between checking your Google rankings once a month and having Search Console running continuously.
What to do with the data
Once you have consistent tracking, the data starts pointing you toward action:
Low appearance rate overall → You're not in the conversation yet. Priority: get mentioned on authoritative sites in your category. Aim for G2, Capterra, Product Hunt, and relevant newsletter features.
High appearance rate on some LLMs, low on others → That LLM's training data or search index has different coverage of your brand. Look at what sources it cites for competitors and get on those.
Appearing late in responses → You're known but not the primary recommendation. Look at competitor positioning — what do they have that you don't in the LLM's framing? Often this is about how the category is defined in high-authority sources.
Good appearance, but competitor growing faster → They're doing something that's earning citation. Check their recent press, comparison content, or community activity.
The founders winning in AI search right now are the ones who started tracking before everyone else. The methodology is simple. The discipline to track consistently is what separates them.