The AI-era metric
Share of Model: the only metric that matters in AI search
How often does AI cite your brand when a buyer asks a question in your category? That number — across ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews — is Share of Model. It replaces keyword rank for the same reason AI is replacing search results lists: there's only one answer slot now. Measuring it is table stakes. Moving it is the job.
Definition
What Share of Model measures
The percentage of AI-generated answers that mention or recommend your brand when the AI is asked a buying-intent prompt in your category. Tested per model, then averaged.
Formula
SoM = ( responses_with_brand_mention
÷ total_buying_intent_prompts )
× 100Composite SoM averages the per-model score across ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews. MentionLayer also tracks a recommended sub-score (positive sentiment + actionable phrasing) and a linked sub-score (your URL appears as a citation).
Why Share of Model replaces keyword rank
One answer, not ten
Google gave 10 blue links per query. AI gives one paragraph. There's no scroll. You're in the answer or you're invisible.
Multiple models, no overlap
The five AI surfaces don't agree — a brand ChatGPT cites every time can be invisible on Gemini, and in our research visibility varies 30× between verticals. Per-model SoM is the only honest read.
Buyer behavior already shifted
65.9% of businesses are invisible to AI (our study of 1,004 businesses). If your category buyers ask AI, your visibility metric is SoM. If they ask Google, it's still keyword rank. Most categories are now both.
Benchmark distribution
What a good Share of Model score looks like
From the AI Visibility Index — our published study of 1,004 businesses.
Invisible
0–5%
65.9% of businesses sit here (n=1,004)
Emerging
5–25%
Citation seeding starts to hit
Recognized
25–60%
Mentioned but rarely the top pick
Category leader
60–95%
Recommended in most answers
Five levers that move Share of Model
Ranked by measured leverage in our published research — not guesswork.
01
Entity & directory sync
Directory presence correlates with AI visibility at r=0.391
02
Cited-source placement
Presence in the sources AI cites shows a 5.5× lift
03
Review velocity
G2, Trustpilot, Google, Capterra
04
Press coverage
Outlets AI cites — not PR-wire churn
05
Technical GEO
Schema, structured data, robots.txt
Share of Model FAQ
What is Share of Model?
Share of Model (SoM) is the percentage of AI-generated answers that mention or recommend your brand when an AI model is asked a buying-intent question in your category. It measures your visibility inside generative AI — ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews — the same way Share of Voice measured your presence inside traditional advertising or SEO. It's calculated per model, then averaged across the five major AI surfaces for a composite score.
How do you calculate Share of Model?
Share of Model = (number of AI responses where your brand is mentioned) ÷ (number of buying-intent prompts tested) × 100. MentionLayer's AI Monitor tests each client against 10–50 buying-intent prompts per week across ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews. The composite SoM is the average across all five surfaces. We also break out 'mentioned only' vs 'recommended' (positive sentiment) vs 'recommended with a link cited' for finer-grained tracking.
Why is Share of Model better than keyword rank?
Keyword rank measures whether you're in a list of 10 blue links on Google. Share of Model measures whether you're in the actual answer the buyer reads. With AI, there's often only one recommendation — the buyer doesn't scroll to the next result. Either you're in the answer or you're not. Share of Model captures that binary in a way keyword rank can't.
What's a good Share of Model score?
Category-leaders typically score 60–85% SoM (mentioned in most AI answers about their category). Strong challengers run 30–60%. Brands new to GEO usually score 0–10%. Per our study of 1,004 businesses (n=1,004), 65.9% are effectively invisible to AI — they're recommended in almost none of the answers buyers see. The gap between invisible and recommended is the largest competitive moat in marketing right now.
How do you increase Share of Model?
Five levers, ranked by the evidence in our research: (1) Directory and entity presence — the strongest off-page predictor we measured, correlating with AI visibility at r=0.391. (2) Getting onto the sources AI answers already cite — businesses present in cited sources show a 5.5× visibility lift. (3) Review velocity on G2/Trustpilot/Google. (4) Press coverage on AI-trusted publications. (5) Technical GEO — schema markup, robots.txt, structured data. Most tools just measure these signals. MentionLayer builds them — and proves the lift against a control group.
How often should you measure Share of Model?
Weekly minimum. AI models update their retrieval and training data continuously, so SoM can shift fast — especially around model releases. MentionLayer's AI Monitor runs every Monday on every active client and posts a delta vs the prior week. Monthly is the cadence we report to clients. Quarterly is the cadence we publish in the AI Visibility Index.
See your Share of Model in about 20 minutes
Free audit tests your brand across ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews. Composite SoM, per-model breakdown, and your top three improvement levers. Then move the number — every plan includes the full execution stack, not just the tracker.