OpenSEOHow we measure visibility
Every metric starts with an observed answer. A missing answer is a gap in coverage, not evidence that a brand is absent.
Prompt language
The language label is supplied by your workspace to describe a prompt. It is not sent as a provider language control and does not establish the language of an answer. Tracked questions are sent unchanged.
Mention rate
The share of successfully analyzed answers that mention your brand in context. Each answer counts once. Answers awaiting analysis are excluded, not counted as missing mentions.
Citation rate
The share of successful answers with an explicit citation to your website or its subdomains. Search results and other attached links are shown separately.
Share of voice
Your brand’s confirmed answer-level mentions divided by the combined confirmed mentions of your brand and selected competitors. Only analyzed answers are included.
Recommendation rate
The share of analyzed answers that recommend your brand. A name appearing in an answer is not automatically a recommendation; analysis must include an exact supporting excerpt.
Coverage
Successful answers divided by scheduled checks. Failed or missing answers reduce coverage; they do not count as answers that omitted your brand.
Comparable history
Changes are compared only across matching prompt versions, countries, prompt-language labels, platforms and competitor sets. Editing these settings changes the comparison group.
Position
A position is shown only when the answer contains an explicit ordered list. A brand appearing earlier in ordinary prose does not establish a ranking.
AI answers vary by location, time and personalization. These checks describe the answers collected under the recorded settings, not every answer every person sees. Incomplete citation extraction is labeled separately. Full answers and reports are retained for 12 months.