The Short Answer
To check if ChatGPT recommends your brand, run the questions your buyers actually type into AI engines and record the results. Build a prompt list of 20 to 40 real buyer questions, run them across ChatGPT and the other major engines, and log whether you are mentioned, whether you are cited as a source, and which competitors appear instead. That prompt set, run consistently over time, gives you a real measure of your AI search visibility. Most brands discover they are less visible than they assumed, and they can see exactly which prompts the competition owns.
The whole check takes an afternoon to set up and gives you a scoreboard you can rerun every few weeks. Everything below is how to run it, what to record, and what to do with the numbers once you have them.
Why You Cannot See This in Google Analytics
Before the method, one thing worth being clear about: your normal analytics will not tell you this. A buyer who reads about you inside ChatGPT and then converts the next day looks like direct or branded traffic, not "AI." There is no referrer, no keyword, no line item that says "recommended by an answer engine." The influence happened inside a conversation you never see.
That is why you have to measure AI visibility directly, by asking the engines the same questions your buyers do and reading what comes back. It feels manual because it is. It is also the only reliable signal you have until you put a tool on it.
The Quick Manual Method
You do not need a tool to start. The manual method takes a few hours to set up and reveals your current position across the engines that matter.
Step 1: Build your prompt list.
Write out 20 to 40 questions your buyers actually ask when they are evaluating options in your category. These are not keyword phrases. They are complete questions that a person would type into ChatGPT: "what is the best [category] tool for [use case]," "who are the top [service] providers for [industry]," "should I use [your brand] or [competitor]," "how do I solve [specific problem]."
Pull these questions from your sales call notes, customer conversations, support tickets, and the queries that bring traffic to your site. If a buyer has asked it out loud, it is a prompt worth testing. Prioritize the questions where a recommendation in the answer would have the most commercial value.
Step 2: Run each prompt across the major engines.
Test at minimum: ChatGPT (both standard and with web search enabled), Perplexity, and Google AI Mode. Add Gemini and Microsoft Copilot if they are relevant to your buyer's behavior. Run each prompt once per engine and copy the full response.
Keep conditions consistent. Use a fresh session for each run or a private window so previous conversation history does not influence the output. Turn off any memory or personalization features, since those quietly tune answers to you rather than to a neutral buyer.
Step 3: Record the results.
For each prompt and engine combination, log four things:
- Was your brand mentioned anywhere in the response? (yes/no)
- Was your brand cited as a source with a link or explicit reference? (yes/no)
- Which competitor brands were mentioned?
- What position was your brand relative to the others?
A simple spreadsheet with prompts as rows and engines as columns works fine at this scale.
Step 4: Run the same set again in two to four weeks.
A single snapshot is interesting. A trend over time is actionable. Patterns that hold across multiple runs tell you which prompts you own and which you are losing.
The Check Process at a Glance
The loop is simple once you see it laid out. Build the prompts, run them, record mentions and citations, score the result, then repeat on a cadence so you are tracking a trend and not a single lucky answer.
What to Track
Once you have a few rounds of data, four metrics tell the story:
Mention rate. What percentage of prompt runs resulted in your brand being named? This is your baseline visibility number. If you run 40 prompts and appear in 8 of them, your mention rate is 20 percent.
Citation rate. Of the responses where you were mentioned, what percentage included a link or explicit source reference to your page? Mentions build awareness. Citations build traffic and model trust. A high mention rate with a low citation rate means the engine knows your name but is not confident enough in your content to point buyers at it.
Position. When you are mentioned, where do you appear? First, second, or third is the consideration set. Fourth or lower is the tail. Average your position across the prompts where you appear. Driving that average up is the medium-term goal.
Share of voice by prompt. For your most commercially valuable prompts, who gets named and how often? If a competitor appears in 80 percent of runs for "best [category] tool for [use case]" and you appear in 20 percent, that prompt is a concrete priority. You can see exactly how much ground you need to gain and in which direction.
The Scoreboard: What Your Sheet Should Look Like
The whole exercise produces one artifact worth keeping: a scoreboard. Each row is a prompt, and the columns record whether you were mentioned, whether you were cited, and who beat you. Here is the shape of it:
| Prompt | Mentioned | Cited | Top competitor |
|---|---|---|---|
| "best [category] tool for [use case]" | Yes | No | Competitor A |
| "who should I hire for [service]" | No | No | Competitor B |
| "is [your brand] worth it" | Yes | Yes | You |
| "top [category] providers for [industry]" | No | No | Competitor A |
| "[your brand] vs [competitor]" | Yes | No | Competitor C |
Read down the columns and the picture is immediate. A column of "No" under Cited means the engines know your category but not your pages. A competitor repeated in the last column is your priority target. A prompt where you are mentioned but never cited is a content and citation gap, not an awareness one. Rerun the sheet every few weeks and the cells that flip from No to Yes are your progress, in a form you can put in front of a stakeholder.
Mentions Versus Citations, and Why the Gap Matters
The single most useful distinction in this whole exercise is mention versus citation, so it is worth slowing down on.
A mention is your brand name appearing in the text of an answer. A citation is the engine linking to or explicitly referencing your page as the source behind a claim. They are independent. You can be mentioned without being cited, which is common, and in rare cases cited without being named in the prose.
The gap between the two tells you where the work is. If your mention rate is decent but your citation rate is near zero, the engines have heard of you but do not trust your own pages enough to send buyers there. And here is the part most brands miss when they read that gap: fixing it is mostly not about your own website.
When we classified every domain AI engines cited for the buyer questions we track, a brand's own site never accounted for more than 6 percent of the citations. The overwhelming majority came from third-party editorial, community, and reference pages. So a low citation rate on your scoreboard rarely means "write more blog posts." It usually means you are absent from the "best of" lists, comparisons, and reviews the engines actually pull from. That is a different, and often faster, fix.
Tools That Automate This
The manual method works at startup. At scale, running 40 prompts across five engines weekly and tracking trends over months requires automation.
The best AEO tools in 2026 handle this loop: Profound for enterprise teams, Peec AI for agencies and multi-engine tracking, and Otterly.ai if you want an affordable starting point. Each tool runs your prompt set on a schedule, records mention and citation rates, and benchmarks you against the competitors you name.
The key consideration when choosing a tool is prompt quality. A tool is only as useful as the prompts you feed it. Platform-generated prompts are generic starting points. The prompts that matter are the ones your actual buyers type, and those require you to bring your own list. The second consideration is whether the tool tracks the source URLs behind answers, not just whether your name appears, since that is what tells you which third-party pages to go win.
What to Do If You Are Not There
If the audit reveals that your brand is invisible or consistently outranked, the fix is not to chase the engine. It is to earn the signals the engine is already rewarding competitors for.
The engines that skip your brand are not doing it arbitrarily. They cite sources that answer questions directly and early, that have structured data making their facts machine-readable, and that are mentioned by credible third parties the model already trusts. Those are the gaps to close.
Practically: rewrite your key pages to answer buyer questions in the first sentence, add schema markup to make those answers extractable, and earn citations on the third-party pages that already rank for your priority prompts. Citation building in particular is underdone by most brands and overrepresented in the results. If you are not named in the "best of" roundups and comparison guides for your category, the engine has no strong signal to cite you.
The measurement work you did above tells you which prompts to prioritize first. Start with the highest-value prompts where you have the smallest gap to close, and build from there. If you want the deeper diagnosis of why an engine skips a brand, why isn't my brand in AI search walks through the usual causes.
Common Mistakes When Checking
A few patterns make a check misleading before it starts.
Running prompts in a logged-in session with memory on, so the engine tunes the answer to you and your brand looks more visible than it is to a neutral buyer. Use a private window.
Testing keyword phrases instead of full questions. "Best [category] tool" behaves differently than the complete sentence a buyer types, and the sentence is what you are actually competing on.
Judging visibility from a single run. Outputs vary, so one lucky answer can flatter you and one unlucky one can panic you. Rates across multiple runs are the truth.
Counting mentions and ignoring citations. A wall of "mentioned: yes" with no citations feels good and changes nothing, because citations are what drive traffic and signal trust.
Checking only ChatGPT. Your buyers also use Perplexity, Gemini, AI Mode, and Copilot, and the answers differ enough that one engine is not a proxy for the rest.
Watch: AEO in Context
If you are new to the discipline behind this measurement, this short explainer from Ahrefs covers what Answer Engine Optimization is and why it matters, which frames why the check above is worth doing.
Key Takeaways
- To check if ChatGPT recommends your brand, run 20 to 40 real buyer questions across the major engines and record mentions, citations, and position versus competitors.
- Track four numbers: mention rate, citation rate, average position, and share of voice on your highest-value prompts.
- A mention is your name in the answer; a citation is a link to your page. The gap between them tells you where the work is.
- Your own website is only 2 to 6 percent of the sources AI engines cite, so a low citation rate usually means you are missing from third-party pages, not that you need more blog posts.
- Rerun the same prompt set on a cadence. A trend is actionable in a way a single snapshot never is.
- The manual method proves the concept; a tool automates it once you are running many prompts across engines weekly.
Where to Start
Do the manual check first. It costs nothing but an afternoon and it will tell you, honestly, whether the engines your buyers use recommend you or hand the answer to a competitor. Build the prompt set from real buyer questions, run it across ChatGPT, Perplexity, and Google AI Mode, and score the mentions and citations. That baseline is the most useful thing you can have before spending on either a tool or an agency.
When you want the full cycle run for you, an AI visibility audit from AEO Labs runs this measurement across your priority prompts, delivers a baseline citation share, and identifies the specific third-party gaps your competitors are filling. From there, our program closes those gaps through content, structured data, and citation building. If you are still mapping the landscape, start with how to rank in ChatGPT or the wider AI search statistics for 2026.