Is My Brand Visible in ChatGPT? How to Test It (Free Template)

Your customers no longer search only on Google. They ask ChatGPT questions like 'Which accounting software should a small business in Ankara use?' and put the three to five brands in the answer on their shortlist. So marketing and sales teams now ask a new question: is my brand visible in ChatGPT, and how do you test it? This guide gives you a repeatable method. It covers how to set up a neutral test, which questions to ask at each stage of the buying journey, how to score the answers, why one test is never enough, and what to do based on the results.

One note before we start: it is hard to find a source that explains this step by step for Turkish brands and Turkish search behavior. This guide aims to fill that gap with a full test protocol.

Short answer: how can you tell in 5 minutes whether your brand shows up in ChatGPT?

Open a temporary chat with memory switched off and turn on web search. Ask 10–20 questions your customers would actually ask, for example 'What is the best dental clinic in Izmir?'. For each answer, note whether your brand is mentioned, its position in the list, whether your website is cited as a source, and whether the details are accurate. Then repeat the test on different days.

5-minute quick check:

1. Open a temporary chat in ChatGPT.
2. Make sure web search is on.
3. Ask 3 real customer questions that do not name your brand.
4. Search each answer for your brand name.
5. Look at the source links in the answer and check whether your website is among them.

This quick check gives you a first signal. The rest of this guide explains how to measure it reliably.

What does 'being visible' in ChatGPT mean? 5 signals to measure

The question 'Did my brand appear?' has more than a yes or no answer. You get a full picture by tracking these five signals:

• Mention: Does your brand name appear in the answer at all?
• Position: Where does it rank in the recommended list? Being first and being sixth carry very different weight.
• Citation: Is your own website linked as a source? A citation sends traffic, and it also shows which page the model treats as a trusted source.
• Tone: Is your brand described positively, neutrally or negatively? Is it 'recommended' or only 'also mentioned'?
• Accuracy: Do your price, address, opening hours and service list match reality?

Asking about your brand vs. asking about the category

This is the most common mistake. If you ask 'What is [Your Brand]?', ChatGPT will usually tell you something about your brand, and many teams stop there with 'Great, it knows us.' Your customers do not ask that way, though. They ask category questions such as 'What is the best pre-accounting software?' The real visibility test is the category question. Brand-name questions are only useful for checking accuracy and tone.

Before the test: how to make ChatGPT neutral

ChatGPT can personalize its answers to you. If you have spent weeks asking about your own brand, the model may remember it, and your results will look better than reality. Prepare the test like this:

• Use a temporary chat: Temporary chats do not use or create memory, and they do not appear in your history. This is the cleanest option for testing.
• Check memory: If you test in a normal chat, turn off memory or saved memories in settings.
• Clear custom instructions: Instructions such as 'I work at company X' or 'I am in marketing' skew the answers. Leave the personalization fields empty during the test.
• Check web search: Answers with web search on are built from sources found at that moment. Answers with search off rely on the model's training data. Test both separately, because the two can give very different results.
• Try another account if you can: A logged-out session or a colleague's account adds a second data point.
• State city and language explicitly: Write location context into the question ('in Ankara', 'in Turkey') and ask in Turkish if your customers search in Turkish.

Which questions should you ask? Sample prompts by buying journey stage

A good test set mirrors how customers actually phrase their questions. Below are sample prompts for three sectors (B2B software, local services/dental clinic, e-commerce). Replace the parts in brackets with your own brand and competitors.

Awareness stage (the customer has not chosen a brand yet)

1. What is pre-accounting software for small businesses, and how do I choose one?
2. What should I look for when choosing a clinic for dental implants?
3. How can I tell whether olive oil I buy online is good quality?
4. What does CRM software bring to a sales team?
5. Which software is used when switching to e-invoicing?

Consideration stage (the customer is building a shortlist)

6. What are the 5 best pre-accounting programs in Istanbul?
7. Which is the best dental clinic in Izmir?
8. Which clinics would you recommend for implants in Ankara?
9. Which are the most trustworthy websites selling cold-pressed olive oil online?
10. What are the most affordable pre-accounting programs for SMEs?
11. Which is better, [Competitor A] or [Competitor B]?
12. What are the differences between [Your Brand] and [Competitor A]?

Decision stage (the customer is checking a specific brand)

13. How much does [Your Brand] pre-accounting software cost?
14. Is [Your Brand] reliable, and what do users say about it?
15. Where is [Clinic Name] located, and which treatments does it offer?
16. How many days does [Your Brand] take to deliver orders, and what is the return policy?
17. What alternatives to [Your Brand] should I consider?

Recommended volume: 10–20 questions, ideally weighted toward the consideration stage, and at least 3 attempts per question. Comparison questions that include competitor names show whether the model puts you in the same tier as them.

Step-by-step test: how do you record and score the answers?

Ask each question in a new temporary chat. Asking several questions in the same chat lets the previous answer influence the next one. Save every answer as a screenshot or as text. Here is what the answers usually look like:

Example view 1, brand mentioned: 'The most popular pre-accounting programs for small businesses in Istanbul are: 1. Brand A, 2. [Your Brand], which stands out with e-invoice integration and an affordable entry plan, 3. Brand C...' Sources: [yourbrand.com], [comparison-site.com]

Example view 2, brand not mentioned: 'Popular options include Brand A, Brand C and Brand D. Brand A is especially preferred by SMEs...' Your name does not appear anywhere in the answer.

Example view 3, source list: At the end of the answer or in the sources panel: a software comparison site, a business news site, Brand A's pricing page, a forum thread. These are the pages the model read before it answered.

Scoring table (template)

Copy the columns below into Google Sheets or Excel and you have a ready-to-use template. Add one row per answer.

Question	Stage	Mentioned (Y/N)	Position	Own site cited?	Tone	Inaccurate info	Competitors mentioned	Source sites
Best 5 pre-accounting programs in Istanbul	Consideration	Y	3	N	Neutral	Old price	Brand A, Brand C	comparison-site, news-site
Best dental clinic in Izmir	Consideration	N	–	N	–	–	Clinic X, Clinic Y	guide-site, maps listing
How much does [Your Brand] cost?	Decision	Y	–	Y	Positive	None	–	yourbrand.com

Two simple formulas

Mention rate = answers where you were mentioned ÷ total answers

Example: You asked 15 questions with 3 attempts each, so 45 answers in total. You were mentioned in 18 of them. Your mention rate is 18 ÷ 45 = 40%.

Share of voice = your brand's mentions ÷ total mentions of all brands you track

Example: Across the same 45 answers, you were mentioned 18 times, Competitor A 27 times, Competitor B 12 times and Competitor C 3 times. Total mentions = 60. Your share of voice is 18 ÷ 60 = 30%, and Competitor A's is 27 ÷ 60 = 45%.

The mention rate shows how often you appear. Share of voice shows how much space you take relative to competitors. Track both together.

Why do I get a different answer every time? Why a single test misleads

If you ask the same question twice, you will often get two different lists. This is not a bug. It is how these systems work:

• Answers are probabilistic: Language models generate text by choosing among likely options. When several brands fit a category equally well, the order and even the list itself can change between attempts.
• Web search sources change: With search on, the answer is built from the pages found at that moment. If the pages found change, the brands in the answer change too.
• Models get updated: When a new model version ships, the same question can produce a different picture.
• Personalization: Memory, earlier chats and location can affect results. That is why the neutral setup above matters.

The fix: Ask each question at least 3 times, record the average rather than a single result, and repeat the test with the same question set every week. A brand that appears in 1 of 3 attempts and a brand that appears in 3 of 3 are in very different positions.

ChatGPT alone is not enough: run the same test in Gemini, Perplexity, Claude and Google AI Overviews

Each answer engine relies on different sources, so being visible in one does not guarantee visibility in another.

Engine	Main source	How to test	Notes
ChatGPT	Web search + training data	Temporary chat, search on/off	Test both modes
Gemini	Google Search	New chat, same questions	Your Google presence has an impact
Perplexity	Its own search and Sonar models	Same questions, check the source list	Sources are shown in nearly every answer
Claude	Web search (when enabled) + training data	New chat, check search settings	Results differ when search is off
Google AI Overviews	Google Search results	Search the question as a normal Google query	Only appears for some queries, languages and regions

To check Google AI Overviews, type your question into Google as a normal search. If an AI summary appears at the top of the page, note whether your brand is mentioned and which links appear on the right or below.

Cross-engine template: Pick 5 core questions, ask them in all 5 engines, and record 'mentioned/position' in each cell. Within a few minutes you can see which engine you are weak in.

Why doesn't ChatGPT show my brand? Diagnosing through source sites

The most valuable data from the test is the source list. It tells you which pages the model read before it answered.

• List the sources: Note which directories, comparison sites, news sites, forums and competitor pages are cited across all answers. Count which domains appear most often.
• Check competitor presence: If a competitor appears on these sites and you do not, that is the most likely reason you are missing. The model recommends what it reads.
• Check your own site: Do you have a page that clearly answers the category question? Pages that state the service, city, price range and frequently asked questions are easier for the model to use. A homepage that only uses slogans is not enough.
• Find the source of inaccurate info: An outdated price or old address usually comes from an external source that was not updated, such as a directory listing, an old news article or a stale comparison table. Find that source, update it or contact the site owner, and make the correct information easy to find on your own site.

This is also why ranking high on Google does not guarantee a ChatGPT recommendation. The model may pull from different pages than the ones you rank with, and it builds its answer from content that answers the question directly.

Manual test or automated tool?

The method above works, but it has limits:

• It takes time: 15 questions × 3 attempts × 5 engines = 225 answers. Recording and scoring that by hand every week is a serious workload.
• Repeatability is low: Different people word the questions slightly differently, which makes comparisons harder.
• It becomes unmanageable: As the number of engines, competitors and questions grows, the spreadsheet breaks.

Automated measurement repeats the same questions on a fixed schedule. It scores results by prompt and by engine, compares you with competitors, and alerts you when visibility drops. For example, Jumpix AI Visibility asks the questions customers ask ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews. It measures whether your brand is mentioned, recommended, cited and described correctly, then lists what to fix. Before you commit, you can try the free AI visibility check that requires no sign-up. It asks 5 buyer questions across 3 engines for your sector and city. A free trial with no card is also available (1 brand, 1 measurement), and for ongoing weekly measurement, AI brand tracking costs 9.99 USD per month for 1 brand. Teams in the Turkish market that prefer not to run this themselves can choose a managed GEO service.

What should you do in the first 30 days based on the results?

Week 1, map content gaps: Group the questions where you were not mentioned by stage. If you are missing at the consideration stage, you need comparison and category pages. If information is wrong at the decision stage, you need clear pricing and FAQ pages.

Week 2, get into source sites: Start with the most-cited domains from your source list. Join relevant directories and comparison lists, update your profile listings, and pursue industry news and content collaborations.

Week 3, make your brand information consistent: Align your price, address, service list and contact details across your website, map listings, social profiles and directories. Contradictory information makes the model less certain.

Week 4, re-measure: Re-run the test with the same question set, the same settings and the same number of attempts. Compare mention rate, share of voice and inaccurate-info count before and after. Visibility does not change overnight, so the goal is a measurement cycle you keep running.

Conclusion

'Is my brand visible in ChatGPT?' cannot be answered with a single question. A sound test needs a neutral setup (temporary chat, no memory, controlled web search) and real customer questions grouped by buying stage. It also needs scoring across five signals (mention, position, citation, tone, accuracy) and enough repetition to account for the probabilistic nature of the answers. Repeating the same test in other answer engines and reading the source lists shows you why you are missing. Do it manually for the first round. Once the process is clear, switch to regular automated measurement so you can see the effect of your changes.

Frequently Asked Questions

How can I check for free whether ChatGPT mentions my brand?

Open a temporary chat in ChatGPT, turn on web search, and ask 10–20 category questions your customers would ask without naming your brand. Note whether your brand is mentioned and where it ranks. Ask each question at least 3 times. If you want a quick automated look, you can also use a free check that needs no sign-up.

Why does ChatGPT recommend different brands every time I ask?

Language model answers are probabilistic. When several brands fit a category, the order and the list can change between attempts. With web search on, the sources found at that moment also change the answer. That is why you should ask each question several times and look at the average.

What should I do if ChatGPT gives wrong information about my brand?

First, check the sources linked in the answer. Wrong information usually comes from an outdated directory, news article or comparison page. Fix those sources or contact their owners, and publish clear, current pricing, address and service information on your own site. Then re-test with the same questions.

I rank high on Google but ChatGPT does not recommend me. Why?

ChatGPT does not copy Google rankings. It builds its answer from the sources it finds and from content that answers the question directly. If competitors appear on the comparison and directory sites the model reads and you do not, or if your site has no page that clearly answers the category question, you can stay out of the answer despite strong Google rankings.

Besides ChatGPT, which AI tools should I test brand visibility in?

Run the same question set in Gemini, Perplexity, Claude and Google AI Overviews. Each engine relies on different sources: Gemini on Google Search and Perplexity on its own search infrastructure, for example. Strong visibility in one engine does not guarantee visibility in another.

Related reading

<!-- rank-agent-related -->

• Why Doesn't ChatGPT Recommend My Brand? 8 Causes and Fixes <!-- rank-agent-link:8d7217c5-cd4c-41e7-9d3f-4f837ec342ae -->
• How to Track Whether ChatGPT Recommends Your Brand: 4-Step Guide <!-- rank-agent-link:28987ea6-fa79-40e4-bae1-4ceafb19274d -->
• What Is ChatGPT SEO? A Guide to Getting Cited by ChatGPT <!-- rank-agent-link:02b602fe-8ecd-48ad-b111-2cee16bedb3d -->