AEO Metrics: Mention Rate, Citation Rate, and Source Quality

AEO metrics should explain how your brand appears inside AI-generated answers.

Do not start with one big score. Start with the signals that create the score: mentions, citations, competitors, source quality, and answer framing.

This guide defines the metrics that belong in an AI visibility score and an AI search visibility report.

Brand Mention Rate

Brand mention rate is the share of tracked answers that mention your brand.

Formula:

brand mention rate = answers mentioning your brand / total tracked answers

Use it to answer:

• Is the brand visible in category questions?
• How often does it appear for comparison questions?
• Brand-only visibility: does the mention happen only when the prompt includes the brand name?
• Channel variance: where does visibility rise or fall?

Do not overvalue branded prompts. If the question includes your brand name, a mention is expected. Non-branded category and comparison prompts are usually more revealing.

Competitor Mention Rate

Competitor mention rate tracks how often alternatives appear in the same answer set.

Formula:

competitor mention rate = answers mentioning tracked competitors / total tracked answers

This metric is useful only when competitor names, aliases, and domains are defined consistently. Otherwise, you will undercount or double-count. Use canonical competitor profiles in the competitor AI visibility workflow.

Interpretation:

• High brand mentions and high competitor mentions: your category is visible, but competitive framing matters.
• Low brand mentions and high competitor mentions: competitors own the answer layer.
• High brand mentions and low competitor mentions: strong visibility, but watch for overfitted query sets.
• Low both: the query set may be too broad, too early, or unsupported by available sources.

Citation Rate

Citation rate measures how often answers cite a source connected to the answer.

Formula:

citation rate = answers with citations / total tracked answers

This is not the same as owned citation rate. An answer may cite review sites, competitor pages, analyst reports, documentation, or no sources at all.

For Perplexity and Google AI features, citations can be central to the user experience. For ChatGPT, citation behavior can vary by product surface and retrieval mode. Preserve source URLs whenever they appear.

Owned Citation Rate

Owned citation rate is the share of answers that cite your own domains.

Formula:

owned citation rate = answers citing your owned domains / total tracked answers

This tells you whether your own pages are being used as evidence.

Owned citations are useful when:

• The cited page is current.
• The answer uses the citation to support the right claim.
• The cited page matches the buyer intent.
• The page is not blocked or canonicalized incorrectly.

Owned citations are weaker when:

• The cited page is outdated.
• The answer cites your page but recommends a competitor.
• The page is thin and only loosely related.
• The citation appears only for branded prompts.

Source Quality

Source quality is a judgment metric, but it should still be structured.

Rate each cited source:

• Strong: current, authoritative, relevant, accessible, and directly supports the answer.
• Usable: relevant but incomplete, narrow, or missing proof.
• Weak: outdated, thin, off-topic, inaccessible, or misleading.
• Risky: competitor-controlled, legally sensitive, or factually wrong.

Google's structured data documentation is useful here because it reinforces a basic principle: markup helps systems understand content, but content quality still matters. A source with schema can still be a weak source.

Source Diversity

Source diversity shows whether answer evidence depends on one type of source.

Track:

• Owned sources.
• Earned media.
• Review sites.
• Analyst or research pages.
• Partner pages.
• Community discussions.
• Competitor pages.

Healthy visibility usually has more than one source type. If all evidence comes from your own site, third-party trust may be thin. If all evidence comes from third parties, your owned pages may not be answering important questions clearly enough.

Answer Framing

Answer framing records how the AI system talks about your brand.

Suggested labels:

• Recommended.
• Listed.
• Compared.
• Mentioned in passing.
• Caveated.
• Absent.
• Incorrect.

Framing turns a mention into a useful signal. A brand can have a high mention rate but weak framing if it appears only as a caveat or late alternative.

Channel Coverage

Channel coverage measures visibility across answer engines.

Track each channel separately:

• ChatGPT.
• Google AI Overview or AI Mode surfaces.
• Perplexity.
• Any other channel your buyers use.

Google explains AI features from a Search owner's perspective in its AI features guidance. OpenAI and Perplexity publish crawler documentation for their systems. These channels do not behave identically, so channel-level reporting matters.

Use the ChatGPT vs Perplexity vs Google AI Overview monitoring guide for the channel breakdown.

Query Coverage

Query coverage checks whether your tracked questions match buyer intent.

Group queries by:

• Discovery.
• Category definition.
• Alternatives.
• Comparisons.
• Pricing.
• Integrations.
• Security.
• Proof.
• Implementation.

If your query set has only broad category questions, you will miss high-intent comparison and proof behavior. If it has only branded prompts, you will overstate visibility.

Metric Review Template

Use this monthly:

The Bottom Line

AEO metrics should help teams decide what to fix.

Start with mention rate, competitor mention rate, citation rate, owned citation rate, source quality, and answer framing. Keep the raw answer evidence. Then roll those signals into a score only after the definitions are stable.

Use AEO Table to track AEO metrics across repeatable Tasks, Runs, citations, competitors, and shareable reports.