How to check whether AI recommends your brand.
Without trusting one answer. Use this small exploratory exercise to record what you actually observed—and what you still don’t know.
No signup required. This is a manual educational worksheet, not an automated report or a validated universal visibility score.
1. Choose five buyer questions
Write realistic questions a potential customer might ask without naming your business. Keep branded checks separate.
- What tools can help [buyer] accomplish [task]?
- Which alternatives to [established product] suit [specific need]?
- What software supports [required capability]?
Choose questions based on actual customer needs where possible. These suggestions are not proven high-volume search queries.
2. Collect twenty observations
Select two relevant AI interfaces. Run each question twice per interface: 5 questions × 2 interfaces × 2 runs = 20 observations.
Record the exact interface or collection method, date, visible search settings and context. API results can differ from consumer chat experiences. Do not submit private customer or business information.
3. Keep the evidence together
Separate a brand mention, an actual recommendation and a cited source. Record missing or failed responses as failures rather than treating them as completed answers without a mention.
| Exact question | Interface / method | Date | Region / context | Brand mentioned? | Recommended? | Competitors | Sources cited | Claims / errors | Next investigation |
|---|---|---|---|---|---|---|---|---|---|
| Observation 1 | — | — | — | — | — | — | — | — | — |
| Observation 2 | — | — | — | — | — | — | — | — | — |
| Observation 3 | — | — | — | — | — | — | — | — | — |
4. Describe patterns with a denominator
If your brand appeared in 6 of 18 completed observations, report exactly that: 6 of 18 completed observations, with two failures if you attempted twenty. This is a hypothetical arithmetic example, not a BrandSeen finding or a probability for all buyers.
Keep question, platform and collection differences visible. A comparison is easier to interpret when you repeat the same method.
5. Choose an evidence-based investigation
- Is the public brand information accurate?
- Is the service or product description clear?
- Do cited third-party pages support the claim?
- Is a potential comparison opportunity relevant to the buyer?
- Is the observation too weak to justify changing anything yet?
Record your reasoning. Updating a page does not guarantee an AI recommendation or establish that a later change was caused by your work.
Want to make this process easier?
BrandSeen is in development. Join early access or apply for the pilot to help shape the experience.
Method and further reading
This is BrandSeen’s educational starting method, not a vendor-endorsed sample-size rule. Google’s AI Search guidance provides context on website discoverability. It does not guarantee inclusion in any AI product.
Written by BrandSeen · 8 November 2026. Revisit the method as interfaces and collection options change.