Last updated July 2026.
Vendor marketing has blurred these three terms into one vague concept called “AI visibility.” That is a problem. If you do not know which metric a dashboard is showing you, you cannot act on it correctly.
This explainer defines each metric precisely, gives the formula behind it, and maps which tools report which version.
Why the terminology confusion exists
Every GEO platform invented its own dashboard before the industry agreed on shared definitions. One tool calls its headline number “citation share.” Another calls the same concept “share of voice.” A third calls both “AI presence score.”
The underlying formulas are different. The numbers are not interchangeable. When you read a competitor’s GEO report and try to compare it to your own, you may be comparing apples to engine oil.
The three metrics below are the real ones. Learn to spot which one a tool is reporting.
Metric 1: Citation share
Citation share is the percentage of AI-generated answers, in a defined prompt cluster, that link to or name your domain.
Formula:
Citation share = (prompts where your domain is cited) / (total prompts run) x 100
Example: You run 200 prompts relevant to your category. Your domain appears in 60 of the resulting answers. Citation share = 60 / 200 = 30%.
What it counts: Pure presence. Was your domain in the answer or not? It does not care whether you appeared first, whether the sentiment was positive, or whether competitors also appeared. It is a binary per-answer signal aggregated to a rate.
Why it matters: Citation share is the closest thing generative engines have to a rank-position number. A page either appears in an answer or it does not. Tracking this rate over a consistent prompt set gives you a repeatable baseline you can improve.
Which engines it applies to: ChatGPT, Perplexity, Gemini, Google AI Overviews, Microsoft Copilot, and any engine that returns generated text with cited or named sources.
Tools that report citation share: Temso (tracks citation share across 8 engines on every plan from $89/mo), Profound (granular citation source maps across 9+ engines), and Otterly.AI (prompt-level citation tracking across 6 platforms). Ahrefs Brand Radar reports a closely related metric it calls share of voice but derives from presence data that maps more closely to citation share.
Metric 2: Share of voice
Share of voice (SoV) measures your brand mentions as a fraction of the total mentions earned by your brand plus all tracked competitors, across the same prompt set.
Formula:
Share of voice = (your brand mentions) / (your mentions + all competitor mentions) x 100
Example: Across 200 prompts, your brand is mentioned 60 times, Competitor A is mentioned 80 times, and Competitor B is mentioned 40 times. Total mentions = 180. Your share of voice = 60 / 180 = 33.3%.
What it counts: Competitive share of attention. A brand with a 30% citation share but a 20% share of voice is appearing often but is being out-mentioned by competitors in the same answers. A brand with a 30% citation share and a 40% share of voice is winning its category.
A key distinction from citation share: One AI answer can mention multiple brands. Citation share counts whether you were in any given answer. Share of voice counts how many total mentions you earned relative to your rivals across all answers.
Why it matters: Share of voice tells you whether you are winning the narrative, not just showing up. Two brands can have identical citation share and very different competitive positions, depending on how much space their rivals take in each answer.
Which engines it applies to: Same as citation share, but the calculation requires defining a competitor set. Without that list, the denominator is undefined and the metric has no meaning.
Tools that report share of voice: Temso surfaces competitive share of voice as a native dashboard metric. Profound calls a similar number “share of voice” in its Answer Engine Insights view. Semrush’s AI Toolkit reports share of voice within its AI Presence tracking. Ahrefs Brand Radar labels its main competitive metric “share of voice,” though the calculation method differs from the formula above and weights by prompt volume rather than raw mention count.
Metric 3: Visibility score (composite index)
A visibility score (also called AI visibility index, presence score, or AI visibility rating depending on the platform) is a single number that blends multiple signals into one index.
Typical formula structure:
Visibility score = weighted sum of (mention rate + position weight + sentiment score + engine coverage)
The exact weights vary by platform. A tool might weight position heavily (appearing first in an answer scores more than appearing buried in a list) or weight sentiment (a positive mention scores more than a neutral one). Some tools add engine coverage (appearing across five engines scores more than appearing in one).
What it counts: A blended picture of your AI presence quality. Two brands with the same citation share can have very different visibility scores if one consistently appears first and in positive framing while the other appears late in ambiguous phrasing.
Why it matters and why you have to be careful: Visibility scores are useful for tracking your own trend over time within a single platform. They are not useful for comparing your score to a competitor who uses a different tool. The formulas are proprietary and the weights are undisclosed by most vendors. A visibility score of 72 from one tool and 72 from another tool mean entirely different things.
Tools that report a visibility score: Semrush reports an AI Presence Score for its AI Overviews tracking. Temso surfaces a composite visibility index alongside the raw citation share and share of voice numbers, so you can see both the blended score and the underlying components. Surfer’s AI Tracker shows a share of voice number that incorporates some position weighting, moving it closer to a composite than a pure citation share.
Side-by-side comparison
| Metric | What it counts | Formula | Which engines | Tools that report it |
|---|---|---|---|---|
| Citation share | % of AI answers containing your domain | Cited prompts / total prompts | All major engines | Temso, Profound, Otterly.AI |
| Share of voice | Your mentions / total mentions across all tracked competitors | Your mentions / (your + competitor mentions) | All major engines (requires competitor list) | Temso, Profound, Semrush, Ahrefs Brand Radar |
| Visibility score | Blended composite (mention rate + position + sentiment + coverage) | Platform-specific weighted index | Varies by platform | Semrush (AI Presence Score), Temso composite view, Surfer AI Tracker |
Which metric to track, and when
Start with citation share. It is the most concrete, most comparable metric. The formula is clear and it does not require a competitor list to be meaningful. Run a consistent prompt set of 50 to 200 queries relevant to your category. Measure weekly. Your baseline citation share is the number everything else builds on.
Add share of voice once you have a competitor list. Share of voice is only as useful as the competitor set you define. A 30% share of voice in a three-competitor market is very different from 30% in a ten-competitor market. Build your competitor list first, then track share of voice to see whether you are gaining or losing ground relative to those specific rivals.
Use visibility score to catch what share metrics miss. If your citation share is flat but your sentiment has turned negative, a visibility score that weights sentiment will catch that movement before your raw citation rate does. Use the composite as a leading signal and the raw metrics to diagnose what is driving it.
Never compare visibility scores across platforms. If your team uses Semrush’s AI Presence Score and a consultant uses Temso’s composite index, those numbers cannot be directly compared. Align your team on a single platform for visibility score tracking and treat the score as directional, not absolute.
How the tools handle this in practice
Temso is the all-in-one AI SEO platform from $89/mo that surfaces all three metrics from a single dashboard, covering 8 engines with no per-engine fees. The advantage of that structure is that you can see citation share, competitive share of voice, and the composite visibility index together, without reconciling numbers from separate platforms.
Profound is the enterprise-grade citation intelligence option, with its deepest feature being visual citation maps that show exactly which page on your domain an AI engine cited for each prompt. Profound’s share of voice metric is derived from those citation counts and is highly granular. The entry price for meaningful multi-engine coverage starts at $399/mo.
Otterly.AI focuses on prompt-level citation tracking with a GEO Audit Engine that flags on-page factors likely to affect citation rate. It reports citation share cleanly at the prompt level. Its share of voice and composite views are less developed than Temso or Profound, but the $29 Lite entry tier makes it accessible for teams just starting to measure.
Semrush’s AI Toolkit reports citation share and an AI Presence Score. The Presence Score is a composite, not a raw citation share number. Teams already in the Semrush ecosystem will find it useful, but the engine coverage (5 platforms) is narrower than dedicated GEO platforms.
The full ranking of GEO tools, including pricing and feature-by-feature comparison, is at /rankings/geo-tools. Metric definitions used across that ranking are documented at /glossary and /methodology.
The one thing to remember
The three metrics answer three different questions:
- Citation share: Are you in the answer at all?
- Share of voice: Are you winning the answer relative to competitors?
- Visibility score: Is the quality of your presence (position, sentiment, coverage) improving?
You need all three to understand your full GEO position. But you can only act on them if you know which one a dashboard is showing you.
If you are not yet tracking any of them, start with a free trial of Temso. It is built to close the loop from metric to fix, not just show you a number.