Presence asks whether a brand, domain or source appears in an answer when the user did not already name it.
A brand listed first is not equivalent to a brand buried after several competitors, even when both are mentioned.
If an engine fails or is not measured, the dataset is incomplete. It should not be converted into a legitimate zero score.
A reproducible AI visibility system should keep every prompt, engine response, extracted signal and score contribution inspectable.
Detect whether the brand or domain is mentioned in non-branded prompts where visibility must be earned.
Record the position of the mention or recommendation so early placements carry more weight than late references.
Separate owned source citations from third-party citations and uncited brand mentions.
Show which engines were measured, failed or missing before interpreting any score.
A model may understand a brand when directly asked about it while still not recommending it spontaneously for category prompts.
Every score needs a path from prompt to engine answer, mention detection, citation detection, position and contribution.
The same panel, prompt grammar, provider list, model IDs and measurement period should be recorded for future comparison.
Timeouts, failed providers and missing observations should remain durable QA evidence instead of disappearing from the dataset.
A citation is evidence that an engine used a source, not proof that the brand owns the answer surface.
A score is most useful when shown with engine coverage, prompt family, citation rate, position and known limitations.
It is the process of testing how often a brand, domain or source appears in AI-generated answers for relevant prompts, then explaining which observations produced the score.
A failed engine says the measurement did not complete. It does not prove the brand was absent, so it should invalidate or qualify the dataset rather than lower the score silently.
Brand understanding tests whether an AI system can explain a named brand. Visibility tests whether the brand appears naturally when the user asks about the category, problem or alternatives.
AI answers vary by prompt and provider. Explainability shows which prompts, engines, mentions, citations and positions created the final result.
Senvarel uses these concepts to structure AI visibility monitoring, free checkers, GEO audits and future public methodology work without exposing confidential implementation details.