Short answer: To monitor brand mentions in ChatGPT, Gemini and Perplexity, ask each AI model the questions your customers really ask — without naming your brand — repeat each question many times, and record how often your brand is named, where it ranks and which websites the AI cites. One check is a guess; repeated checks are a measurement.
Key takeaways
- Test unaided questions (no brand name) — they show whether AI recommends you on its own.
- Ask every question many times. Pranas asks each question 20 times per AI Scan because answers vary.
- Track four numbers: Mention Rate, Coverage, average rank and AI Citations.
- Across 88 brands, the average Mention Rate was 28.3% — but it ranges from 9.6% to 57.8% by industry.
- Monitor each AI model separately; ChatGPT, Gemini and Perplexity do not agree.
Why AI brand monitoring is different from social listening
AI brand monitoring measures what AI assistants say when someone asks for a recommendation. Social listening tracks what people post. The difference matters because an AI answer is generated fresh each time: there is no public post to find, and two people asking the same question can see different brands.
That means you cannot "search" for your AI mentions. You have to ask the questions yourself, repeatedly, and measure the pattern.
Step 1: Choose the questions your customers ask
Start with 10–30 real questions. Good sources are your search console queries, sales calls and support tickets.
- Category questions: "What is the best mobile plan for students?"
- Problem questions: "How do I fix oily skin in humid weather?"
- Comparison questions: "Brand A vs Brand B for business banking?"
Keep most questions unaided (no brand name). Unaided questions show whether AI chooses you; aided questions only show what AI says once you are named.
Step 2: Pick the AI models to monitor
Monitor the assistants your audience actually uses. For most brands that means ChatGPT, Gemini, Google AI Overview and Perplexity, plus Claude or DeepSeek in some markets. Pranas runs AI Scans on 19 AI models from 8 providers so you can compare them side by side.
Step 3: Ask each question many times
A single answer is not a measurement. AI models produce different answers on each run, especially when they search the web. Pranas asks each question 20 times per AI Scan and reports the share of answers that named your brand.
Step 4: Measure the four core numbers
| Metric | What it tells you | How it is measured |
|---|---|---|
| Mention Rate | How often AI names your brand | Share of answers that mention you |
| Coverage | How widely you show up across questions | Share of AI Scans where you appear at least once |
| Average rank | How high you are listed when named | Your average position in the answer |
| AI Citations | Which websites AI relies on | Domains cited in the answers |
Pranas combines these into an AI Score so you can track one headline number over time. See AI search visibility metrics for full definitions.
Step 5: Compare against your industry
A 20% Mention Rate can be strong or weak depending on your category. In Pranas data from 3,519 unaided AI Scans:
| Industry | Brands | AI Scans | Mention Rate | Coverage |
|---|---|---|---|---|
| Financial services | 11 | 295 | 57.8% | 74.2% |
| Food & beverage | 9 | 355 | 52.6% | 73.0% |
| Automotive | 10 | 138 | 41.8% | 55.1% |
| Telecommunications | 9 | 1,372 | 14.2% | 36.9% |
| Beauty | 10 | 335 | 9.6% | 23.9% |
| All brands | 88 | 3,519 | 28.3% | 48.7% |
How we measured. 3,519 completed unaided AI Scans for 88 brands across 5 industries, run between 17 July and 17 October 2026. Unaided means the question did not name the brand. Each AI Scan asks one question 20 times. Mention Rate is the average share of those answers that name the brand. Coverage is the share of AI Scans in which the brand appears at least once. Data is anonymised and aggregated; every industry includes at least 5 brands, and Pranas's own brand is excluded.
Read more in AI visibility benchmarks by industry.
Step 6: Repeat on a schedule and act on the gaps
Run the same questions weekly or monthly so you can see trends, not snapshots. When a competitor appears and you do not, look at the AI Citations for that answer: the cited websites are where AI learned about your competitor. See how AI models choose their sources.
Manual checks vs a monitoring tool
| Manual checks | Pranas AI Scans | |
|---|---|---|
| Answers per question | 1–3 | 20 |
| Models | One at a time | 19 models side by side |
| Metrics | Notes in a sheet | Mention Rate, Coverage, rank, AI Citations, AI Score |
| History | Hard to keep | Saved and scheduled |
Manual checks are fine for a first look. For decisions, use repeated sampling.
Frequently asked questions
How do I monitor brand mentions in ChatGPT? Ask ChatGPT your customers' questions without naming your brand, repeat each one many times, and record Mention Rate, position and sources.
Can I see brand mentions in Gemini and Google AI Overview? Yes — test them the same way, and monitor each model separately.
How often should I check? Weekly or monthly. AI answers shift when models update.
What is a good Mention Rate? The average across 88 brands was 28.3%, from 9.6% (beauty) to 57.8% (financial services).
Next step: Run a free AI Scan to see your brand's Mention Rate in ChatGPT, Gemini and Perplexity. Related: AI brand monitoring · Monitor brand mentions · Pricing
About the author. The Pranas Research Team analyses thousands of AI Scans across ChatGPT, Gemini, Google AI Overview, Claude, Perplexity and DeepSeek to measure how AI search represents brands.