How to check whether ChatGPT mentions your brand.
You do not need a mysterious “AI ranking score” to start. Build a small, repeatable set of buyer questions, record what each answer actually says, and keep the prompt, model, date, and evidence alongside every observation.
A practical measurement framework from CiteGlance · October 2026
1. Start with buyer questions, not vanity prompts
Write prompts that reflect real decisions in your category. Include broad discovery questions, use-case questions, comparisons, alternatives, and questions about trust. Avoid only asking “What do you know about [brand]?”—that tests recall, not whether a buyer would encounter your brand naturally.
Category discovery
What are the best [category] tools for a [company type]?
Use-case fit
Which [category] product is best for [specific use case]?
Comparison
Compare [your brand] with [competitor] for [buyer need].
Alternatives
What are alternatives to [competitor] for a [team size] team?
Trust and evidence
Which [category] providers have strong evidence for [important outcome]?
2. Record the answer and its evidence
For each run, save the exact prompt, platform and model when known, date, market or locale, whether your brand appeared, whether the description was accurate, and any source URLs shown. Keep the original answer or a dated capture where permitted. Do not treat an unlinked brand mention as a citation.
Separate at least three outcomes: mention (the brand is named), citation (a source URL is provided), and recommendation fit (the answer connects the brand to the buyer's stated need). They are related but not interchangeable.
3. Compare competitors fairly
Use the same prompt set for your brand and competitors. Count observed appearances across the test set, note which URLs are cited, and classify the sources: your own site, independent editorial coverage, review platforms, directories, or other sources. Small prompt sets are directional evidence—not a universal market-share estimate.
Repeat the test on a consistent schedule. Answers can vary by account, region, prompt wording, retrieval availability, and model updates, so preserve the conditions and label any changes you cannot control.
4. Turn findings into changes you can defend
Fix factual gaps
Make your product, audience, geography, differentiators, pricing model, and evidence clear and consistent across public pages.
Improve answer-ready pages
Answer specific buyer questions directly, cite primary evidence, show dates where freshness matters, and make authorship and company identity clear.
Understand cited sources
If independent pages repeatedly appear, study what they cover and pursue legitimate inclusion or corrections. Do not manufacture reviews or spam directories.
Re-test before claiming impact
Run the same prompts again after changes. Report observed differences and limitations; do not attribute every change to a single edit without evidence.
Can this be fully automated?
Parts can be automated, but the measurement method matters. Approved model APIs can help run a repeatable prompt set; they are not necessarily identical to the consumer-facing ChatGPT search experience. Reliable tracking should state which product, endpoint, account type, and retrieval mode were actually tested. A technical website scanner alone cannot tell you whether ChatGPT recommends your brand.
For a useful business baseline, combine actual answer observations with technical checks, citation research, and—when access is available—qualified visits and lead data. Treat scores as internal tracking aids, not promises of ranking or revenue.
Find technical gaps before changing content.
Run the free homepage check for metadata, crawlability signals, structured-data presence, and standard discovery files. It does not query AI engines or claim to measure brand mentions.
Run the free AEO checkWant help designing a repeatable prompt set and interpreting evidence? Request an AEO baseline.