In this guide
  1. Define visibility before counting it
  2. Choose a question sample you can repeat
  3. Keep an observation log
  4. Separate external visibility from visits
  5. Prioritize what the source pages are missing
  6. Connect the report to a business decision

Define visibility before counting it

AI visibility is an observation about whether and how information appears in AI-assisted discovery. It is not a single business outcome. Decide whether you are tracking mentions, linked citations, descriptions of your service, or another clearly defined event.

Write the definition before collecting data. If one person counts a brand mention and another counts only a linked source, their totals cannot be compared meaningfully.

Choose a question sample you can repeat

Start with questions a plausible customer would ask. Include informational and service-selection questions, and keep the sample manageable enough to review carefully. Record the exact wording instead of paraphrasing it later.

Document the systems, modes, locations, and settings used. If a setting is unknown, record that limitation. Do not present a convenience sample as a measurement of every possible customer search.

Keep an observation log

A useful log preserves the output and its context. Capture only what you are permitted to retain and avoid including personal account information in shared reports.

Illustrative measurement log structure; fill with actual observations
FieldPurpose
Date and exact questionMakes the check repeatable
System and settingsDefines the observation conditions
Brand mentionRecords whether the name appeared
Linked sourceRecords the exact citation URL, if any
Accuracy issueSeparates presence from correctness
Website referralRecords a separate analytics event when available
LimitationExplains missing context and sample boundaries

Separate external visibility from visits

A person can see a business name without visiting the website. A source can be linked without receiving a click. Use website reporting to understand visits where that data is available, and acknowledge where attribution is incomplete.

Do not fill missing attribution with assumptions. A direct visit may have many possible origins. Keep observed referrals separate from a modeled or inferred explanation.

Prioritize what the source pages are missing

Review inaccurate descriptions and missing context before chasing a larger mention count. A clear statement of the service, audience, process, and limitations helps people even if a generated answer never changes.

Record the exact page improvement and the date it was made. That record supports a useful investigation later; it does not by itself establish that the change caused an external result.

Connect the report to a business decision

Qualified enquiries depend on your own fit criteria. Define those criteria and how the team records the outcome before connecting visibility reports to revenue discussions.

Use the combined evidence to decide what to inspect next. A useful report ends with a question or action grounded in the observations, not a guaranteed growth forecast.