Skip to main content

What this is

A single source of truth for how Meridian labels and interprets metrics.

Why it matters

  • Prevents “we’re arguing about definitions” during reporting.
  • Makes KPI movement actionable (not confusing).
  • Ensures teams interpret results consistently.

Where to find it

Reference → Metrics glossary

Core metrics

Visibility (%)

How often you are present in AI answers across your tracked prompts.
  • Higher is better.
  • Visibility can increase even if rank worsens (you’re included more often, but lower).

Citation Rate (%)

How often answers that include your brand/product also include citations (sources).
  • Higher usually means more trust and more stable rankings.
  • Low citation rate often means you’re “named” but not “sourced.”

Sentiment (%)

A composite score of how positively AI describes your brand across dimensions.
  • Higher is better.
  • Dimension scores (e.g., Fees) may matter more than the overall score for ranking.

Prominence (rank)

Average position when you are mentioned (lower number is better).
  • Example: #3.8 means you’re typically around 3rd–4th place.
  • Prominence can worsen even if visibility stays flat.

Prompt-level metrics

Mentioned (Yes/No)

Whether your brand/product appears in a specific response.

Position (#)

Your rank in the list of mentions for that response (e.g., #1, #2).

Share of Voice (%)

How much of the “recommendation space” you own relative to competitors across tracked prompts.

Citations breakdown

  • Owned: your domain(s)
  • Off-page: third-party editorial sources
  • Competitor: competitor domains
  • Social: forums/social platforms

How to interpret results (quick rules)

  • If Visibility is high but Prominence is poor → you’re present but not preferred → strengthen proof + structure.
  • If Citation Rate is low → trust gap → improve citeability (owned pages + off-page mentions).
  • If sentiment drops in one dimension → narrative gap → add explicit factual sections and FAQs.
  • If one platform differs → platform ecosystem mismatch → optimize by platform.