What this tool actually does
You give us a brand name and a question a real customer might ask. We send that question to ChatGPT (gpt-4o by default), then scan the response for your brand and the brands it mentions instead. The result is a single, honest answer: does ChatGPT recommend you for this query, and who beats you when it doesn't?
This is the most important data point in AI visibility. Search rankings tell you who appears in the top 10 results. AI mention rate tells you who appears in the answer - the only thing the user actually reads.
Why ChatGPT mentions matter more than rankings
Roughly 200 million people use ChatGPT every week. When they ask "what's the best CRM for a 5-person startup?", they get a paragraph naming three or four companies. They do not get a list of ten blue links to scroll through. They make their shortlist from the first answer.
If your brand isn't in that paragraph, you don't exist in the funnel. This tool tells you whether you're in or out, instantly, for any question you can think to ask.
How to read your result
- Mentioned - your brand name appears in ChatGPT's answer. We highlight the surrounding sentence so you can see the context (positive, neutral, or qualified).
- Not mentioned - the response named other brands instead. The competitor list shows who took the slot you wanted.
- Competitors - capitalised brand-style terms ChatGPT named in the same answer. These are the rivals the AI considers comparable to you.
What to do next
- If you were mentioned: run the same prompt 10 more times (rephrased) to see if mention rate is consistent or fragile.
- If you were not mentioned: study the brands that were. What do their pricing pages, reviews, and citation profile look like? AI models pull from third-party signals more than from your own marketing copy.
- Strengthen your brand mentions on the third-party sites that ChatGPT trusts (review sites, comparison roundups, niche communities, Wikipedia where appropriate).
- Make sure your own pages are crawlable by AI bots and that you have a clean llms.txt.
- Move from one-off checks to continuous tracking so you can see the score change as you iterate.
The first time founders run this they always run it on their best-known prompt. They get a result, smile, close the tab. The interesting question is what happens on the prompt where they are NOT the obvious answer. That's where the strategy work begins.
How ChatGPT decides which brands to mention
ChatGPT's answer is generated probabilistically from training data, RLHF, and (in browsing mode) live retrieval. There is no single "mention list" the model consults. Instead, brand inclusion is a function of three overlapping signals.
1. Citation density in training data
Brands that appear frequently across the open web - particularly on third-party review sites, comparison roundups, news outlets and high-authority forums - are over-represented in the model's implicit category map. This is why "ranks well in G2, Capterra, niche subreddits" correlates strongly with AI visibility.
2. Co-occurrence with category language
When the model sees "CRM" and "HubSpot" appear in the same paragraph 50,000 times, it associates them. Co-occurrence with category-defining terms (the words customers use, not necessarily the words you use) is the second-strongest signal.
3. Live retrieval signals
In browsing/search mode, ChatGPT performs a real search and weighs the live results. Standard SEO levers apply: sites that rank well for the query, have schema markup, and load quickly are more likely to be cited.
Mention rate benchmarks by category
Useful sanity checks for what "good" looks like. These are rough averages from our cross-customer data; treat them as a directional guide, not a target.
| Category | Top-1 brand mention rate | Top-5 brand mention rate |
|---|---|---|
| Mature SaaS (CRM, helpdesk) | 30-50% | 10-25% |
| Emerging SaaS (AI tooling) | 15-30% | 5-15% |
| Local services | 20-40% | 5-12% |
| Ecommerce DTC brands | 10-25% | 3-10% |
| Niche / specialist tools | 5-20% | 2-8% |
Improving your ChatGPT mention rate: a 5-step playbook
- Pick the 10 prompts that matter. Not the prompts where you wish customers asked - the ones they actually ask. Sales transcripts and support tickets are the goldmine.
- Run a baseline. Sample each prompt 30 times across all five major engines. Aggregate to a single mention-rate number per prompt.
- Audit citation profiles. Identify the third-party sites that cite the brands ChatGPT prefers. Those are the sites you need to be on.
- Ship one citation a week. A G2 listing, a roundup inclusion, a Reddit thread. Earned coverage compounds; ads do not.
- Measure weekly. Mention rate moves slowly but it moves. A 5-percentage-point lift over six weeks is a real win.
ChatGPT vs Perplexity vs Gemini: which to prioritise
| Engine | Audience type | Citation style | Update cadence |
|---|---|---|---|
| ChatGPT | Mass market, all verticals | Inline mentions, sometimes citations in browsing mode | Weekly to monthly |
| Perplexity | Researchers, B2B buyers | Always cited, footnoted answers | Daily; live retrieval |
| Claude | Developers, knowledge workers | Mentions; citations in Claude-Web mode | Weekly |
| Gemini / AI Overviews | Search-intent users | Inline links to ranking pages | Continuous (Google-driven) |
| Grok | X / social audiences | Mentions; less citation discipline | Daily |
Common mistakes when interpreting results
- Drawing conclusions from one check. LLMs are stochastic. The same prompt can mention you on run 1 and skip you on run 2. Always sample.
- Optimising for prompts your customers do not ask. "Best AI tool 2026" is too broad to convert. Specific intent prompts ("best AI tool to track competitor pricing for a 5-person team under $100/mo") are where mention rate matters.
- Confusing "mentioned" with "recommended". Being named in a long list is not the same as being the headline answer. The paid product distinguishes between top-3 and tail mentions.
- Ignoring sentiment. A negative mention can be worse than no mention. ChatGPT does occasionally surface concerns; the paid product runs sentiment analysis on every result.