3 min read

What metrics measure AI search visibility?

AI visibility metrics show how often assistants name your brand, link your pages, and describe you correctly. Learn mention rate, citation rate, competitor gaps, and snapshot coverage with a worked example.

AI visibility metrics are the numbers that show how often AI assistants name your brand, link your pages, and describe you correctly for the buyer questions you care about. Start with four: mention rate, citation rate, competitor gaps, and snapshot coverage. Traffic and conversion signals come after, as supporting evidence.

Mention rate: how often you're named

A mention is the assistant naming your brand in its answer with no link attached. Mention rate is the share of tracked answers that name you. A common formula is mentions divided by total responses tracked.

Assistants name brands more often than they link them. A citation-only view can make you look invisible on a surface where you're actually being recommended.

Track mention rate for high-intent, decision-stage questions, and watch the trend over time instead of relying on one-off spot checks.

Citation rate: how often your pages are linked

A citation is the assistant linking your specific URL as a source. Citation rate is the share of tracked answers that link your domain.

Note whether you appear as a primary source or a secondary mention, and watch citation momentum over time too.

RootSmith tracks citation rates alongside mentions, so the two numbers stay separate.

Competitor gaps: who gets named instead of you

A competitor gap is a tracked question where a rival is named or cited and you aren't.

Share of voice compares your mentions with competitors' mentions across all tracked responses. Citation share does the same for links within the same answer set.

RootSmith tracks competitor mentions in the answers to your tracked buyer questions. That turns the gap list into a to-do list for new or refreshed pages.

Snapshot coverage: how much of the picture you've sampled

Snapshot coverage is how many tracked questions you've checked, on how many engines, and how many times each. It sets how far you can trust the other rates.

The same prompt asked twice often gives two different answers, so one run proves little. Sample each prompt several times per engine in a logged-out session, and record the rate, not the outcome.

Ask what sits in the denominator of any score. A score you can't reconstruct is a score you can't act on.

For a starting set, list 20 to 30 non-branded questions about problems, solutions, and comparisons that your buyers ask.

Worked example: one question across five engines

Take one tracked question, such as which tool a buyer should pick in your category. RootSmith checks what ChatGPT, Gemini, Perplexity, Claude, and Google AI Overviews say, daily or weekly.

Here is an illustrative reading with made-up numbers:

  • You're named in 3 of 5 engines: mention rate 60%.
  • You're linked in 1 of 5: citation rate 20%.
  • A competitor is named in 4 of 5: that's your gap.
  • Snapshot coverage is 5 engines for this question, with a single run each. That's thin, so repeat each check several times before trusting the rates.

The gap between 60% and 20% suggests buyers may know your brand while your pages aren't being used as sources. The competitor gap tells you which answers to study first.

Repeat this across the full question set, and compare the same set over time.

Supporting signals that sit beside the core four

Accuracy and framing matter too. Check whether descriptions of your brand match your positioning, and watch for misinformation or oversimplified claims.

For traffic and outcomes, look at AI-referred sessions, assisted conversions, and branded search lift. AI referral counts are a floor, because some visits arrive without a referrer and land under Direct in GA4.

Expect directional numbers. Sampling methods vary by tracker, so compare a tool only against its own history.

Measuring what changes after you publish

RootSmith checks each published page's AI citations and Google clicks at 14, 30, 60, and 90 days. That ties page-level movement to the metrics above.

Because RootSmith re-checks the answers after publishing, mention rate, citation rate, and gaps become a loop: find the gap, publish a source-checked page, measure again.

See what AI says about you

Start free. Pick a plan when you're ready for tracking and content.