AI visibility is a strategic opportunity, but to get budget and resources you need KPIs that leadership understands and can act on. Share of Model (SoM) is the top-level metric — your share of relevant AI recommendations compared to competitors. But SoM alone is not enough. You need a complete framework of KPIs covering the entire path from visibility to business value.
What is Share of Model?
Share of Model measures what proportion of relevant AI answers include your brand. The calculation is simple: take a set of questions your ideal customers ask AI models, run them systematically, and count how often you're mentioned compared to competitors.
Example: You define 50 relevant prompts for your industry. ChatGPT mentions you in 12 answers, Competitor A in 22, and Competitor B in 8. Your SoM is 12/(12+22+8) = 28.6%. Competitor A leads with 52.4%.
The five core KPIs for AI visibility
Share of Model is the top-level metric, but underneath it you need five operational KPIs:
- Citation Rate — Absolute number of times your brand is cited in AI answers per period. Shows volume independent of competitors. Trend over time is more important than single values.
- Platform Coverage — Percentage of AI platforms where you're cited. Are you cited on ChatGPT, Gemini, Perplexity, and Google AI Overviews, or just one? Broad coverage reduces platform risk.
- Sentiment Score — Qualitative assessment of the mention. Is it positive, neutral, or negative? A citation with caveats counts against you. Measured on a scale from -1 to +1.
- Entity Recognition — Degree of correctness in the AI model's understanding of your brand. Right category, correct products, right information. Measured by asking identification questions and evaluating response precision.
- AI-Referred Conversion Rate — Conversion rate for visitors coming from AI platforms compared to other channels. Connects AI visibility directly to business value.
How to build a KPI dashboard
An effective AI visibility dashboard should have three levels:
- Leadership level — Share of Model (trend), AI-referred traffic (trend), and estimated business value. Max 3–4 numbers on one slide.
- Strategic level — Citation frequency per platform, sentiment score, entity recognition, and competitor benchmark. Updated monthly.
- Operational level — Citation per prompt, per platform, per period. Which prompts generate citations? Which don't? Updated weekly.
From KPI to action with the decision matrix
- Low SoM + low citation rate — Fundamental problem with entity recognition. Start with Entity SEO and answer engine optimization (AEO).
- Low SoM + high citation on some prompts — You're cited but only in a narrow niche. Expand content coverage.
- High SoM + low sentiment — AI knows you but mentions you negatively. Prioritize reputation management.
- High SoM + low conversion rate — Visibility is there but it doesn't convert. Optimize landing pages for AI-referred traffic.
- Declining SoM over time — Competitors are taking market share. Analyze what they're doing differently.
How to set Share of Model targets
- Starting point — Establish baseline with minimum 30 prompts over 4 weeks.
- Short-term goal (3 months) — Increase SoM by 5–10 percentage points through Entity SEO and content production.
- Medium-term goal (6 months) — Reach parity with closest competitor on at least 2 of 4 platforms.
- Long-term goal (12 months) — Market leader position in SoM for your core categories.
CitationLab and automated KPI tracking
CitationLab makes this entire KPI framework operational. You define prompts and competitors, and the platform delivers automated dashboards with SoM, citation frequency, sentiment, platform coverage, and entity recognition. Reports can be exported for leadership presentations.
Frequently asked questions about Share of Model KPIs
How many prompts do we need for reliable SoM? Minimum 30 for a meaningful baseline. Ideally 50–100 covering the entire buying funnel. Prompt quality matters more than quantity.
How often should we report? Monthly for leadership, weekly for operational teams. AI answers change relatively quickly, so more frequent measurement provides better trend data.
Frequently asked questions
What is Share of Model (SoM)?
How do you calculate Share of Model?
How many prompts do you need for a reliable SoM?
Which AI visibility KPIs should leadership see?
Definitions used in this article
- Share of Model (SoM)
- Share of Model (SoM) is the proportion of relevant AI answers that include your brand, relative to competitors. It is the top-level metric for AI visibility, calculated by running a set of customer prompts and counting mentions.
- Citation Rate
- Citation Rate is the absolute number of times a brand is cited in AI answers per period. It shows volume independent of competitors, where the trend over time matters more than single values.
- Platform Coverage
- Platform Coverage is the percentage of AI platforms — such as ChatGPT, Gemini, Perplexity and Google AI Overviews — where a brand is cited. Broad coverage reduces dependence on any single platform.
- Sentiment Score
- Sentiment Score is a qualitative measure of whether an AI mention is positive, neutral or negative, typically on a scale from -1 to +1.
- Entity Recognition
- Entity Recognition is the degree of correctness in an AI model's understanding of a brand — the right category, products and information — measured by asking identification questions and evaluating the precision of the answers.