Notra: Build an AI Visibility Baseline You Can Interpret

Choose questions with a real constraint

Instead of only asking for “the best project app,” include the audience, task and constraint: “What can a two-person studio use to track client revisions without asking clients to create an account?” Use constraints your product actually addresses.

Keep branded questions separate from category discovery. An assistant describing a product it was explicitly given is a different result from mentioning it in an unbranded comparison.

1. Record each prompt, engine, language and scan date.
2. Keep an initial set unchanged while you establish a baseline.
3. Save the answer and sources, including answers that omit your product.

Read the evidence behind a mention

A mention can be inaccurate, conditional or unfavorable. Read the surrounding sentence and open the cited page. Check whether pricing, eligibility and platform details are current before treating the answer as useful visibility.

Notra also offers conversations that continue a buyer’s questions. Use a follow-up such as “Which option works without installing software?” to see whether the recommendation remains relevant under the same real constraint.

Keep three different measurements separate

A prompt result records an answer to your test question. A crawler request records a fetch. A referral records a person arriving from another service. None, by itself, proves a sale or guarantees a future recommendation.

Notra documents traffic collection and optional provider zero-data-retention settings. Review the integration and retention choices before connecting it. If you change a product page afterward, record the change and compare repeated observations; do not attribute one different AI response to the edit with certainty.