INSIGHT · METHODOLOGY
From question sets to human review:
how to read an AI visibility baseline.
A baseline is useful when its denominator, observation conditions, exceptions and review state are visible. A single combined score can hide the information needed for action.
DIRECT ANSWER · CURRENT SCOPE
What makes an AI visibility baseline usable?
A usable baseline discloses its denominator, question sets, market, language, platform, dates and data sources. It preserves raw answers, cited URLs, no-answer states, errors, unknown sources and off-topic responses. When change matters, observations should be repeated on different dates under the same locked conditions, followed by human review of question intent, brand identity, citation status and completeness.
Boundary: A combined score can hide exceptions and denominator changes. A baseline supports prioritisation; it does not predict future AI answers.
Define the question set first
Separate non-brand buyer questions from branded diagnostics and competitor comparisons. Lock the market, language, surface and question version before observation.
Preserve every state
Keep raw answers, cited URLs, timestamps, no-answer states, errors, source unknowns and off-topic responses. Exceptions remain evidence but stay outside the core denominator until reviewed.
Repeat under comparable conditions
Where a trend matters, observe on multiple dates using the same documented configuration. A changed answer can be a signal, not proof of causation.
Human review is a gate
Reviewers verify question intent, brand and competitor identity, citation status and completeness. AI-generated classifications remain hypotheses until accepted.
A visibility baseline supports prioritisation. It is not a platform ranking and does not guarantee future answers.