New MetricUpdated May 2026

Share of Model Score:
The New AI Visibility Metric

Share of Voice measures your presence in search results. Share of Model measures your presence inside AI-generated answers. In 2026, the second metric is growing faster — and most brands aren't measuring it at all.

By Wahid Ryland, Cited10 min read5 platforms · Complete measurement framework
<6%
Average B2B brand SOM in AI search
Cited Benchmark 2026
14.8x
Sentiment gap between Perplexity and ChatGPT for same brands
Superlines 2026
59.8%
Google AI Overview citations going to brand domains
Otterly 2026
2028
Year AI search projected to surpass traditional search
Siftly 2026

TL;DR

  • Share of Model (SOM) = how often your brand is cited in AI answers ÷ total category queries × 100
  • SOM must be measured per platform — only 11% of domains overlap between ChatGPT and Perplexity
  • Most B2B brands have under 6% SOM — the starting point for a GEO program
  • AI referral traffic converts at 15.9% (ChatGPT) vs 1.76% Google organic — SOM drives high-quality leads
  • The organic AI visibility window is closing — ChatGPT and Google are both testing paid AI placement
  • Brands that build SOM now compound advantage — citation authority accumulates like domain authority

What is Share of Model (SOM)?

DEFINITION

Share of Model (SOM) is a GEO metric that measures how often your brand is cited in AI-generated answers when users ask questions in your category. It is the AI search equivalent of Share of Voice.

SOM Formula:

SOM = (Responses citing your brand ÷ Total queries tested) × 100

Share of Model is measured per platform — your ChatGPT SOM, Perplexity SOM, Gemini SOM, and Claude SOM are tracked separately because only 11% of domains are cited by both ChatGPT and Perplexity. A brand can have strong SOM on one platform and zero on another.

The metric captures both citations (when AI links to your domain) and mentions (when AI names your brand without linking). Both are tracked separately. Citations generate trackable referral traffic. Mentions build brand association in the buyer's mind without generating clicks — a growing share of AI influence happens entirely in the dark funnel.

SOM is a leading indicator. Citations rise first — traffic and pipeline follow, typically 4–8 weeks later. A brand monitoring only GA4 will systematically undercount AI-driven influence because ChatGPT frequently mentions brands without linking.

Share of Model vs Share of Voice

DimensionShare of Voice (SOV)Share of Model (SOM)
Where measuredSearch results page (SERP)Inside AI-generated answer text
FormatRanked list of blue linksEmbedded citation or mention in prose
User interactionUser clicks through to your siteUser reads your brand in the answer — may not click
Primary platformGoogleChatGPT, Perplexity, Gemini, Claude, AI Overviews
Tracked inSEMrush, Ahrefs, Google Search ConsoleManual audits, Otterly, Siftly, Superlines
Conversion rate1.76% (Google organic average)15.9% (ChatGPT referral) — 9x higher
Dark funnel?No — click-based attributionYes — mentions without links are invisible to GA4
Growing or declining?Declining for informational queriesGrowing — 527% YoY AI referral session growth

SOM doesn't replace SOV — it sits alongside it. The brands winning in 2026 track both and invest in both. SOV covers Google; SOM covers the rest.

SOM Benchmarks by Tier

Where does your brand sit? These benchmarks are based on Cited's 2026 analysis of B2B brand citation rates across ChatGPT, Perplexity, and Gemini.

0–6%
Below average

Starting point for most B2B brands. AI tools barely recognize the brand. Immediate GEO work required.

7–15%
Developing

GEO work is showing early results. Brand appears in AI answers but inconsistently. Continue building.

16–30%
Strong

Recognized category presence. AI tools regularly cite the brand. Competitive advantage is forming.

31–50%
Dominant

Category authority. AI tools default to citing this brand. Significant competitive moat.

50%+
Category ownership

Brand is synonymous with the category in AI answers. Compounds over time.

Realistic target: Most B2B brands starting a full GEO program should expect to reach 15–25% SOM on Perplexity within 90 days, and 10–20% on ChatGPT within 120 days. Gemini and Claude vary by industry.

How to Calculate Your SOM

You don't need a paid tool to start. Here is the complete manual SOM calculation method used by Cited for every client audit.

1

Build your query set (20–50 queries)

Write the prompts your buyers actually use. Include 4 query types: Category queries ('what are the best [your category] companies?'), Problem queries ('how do I solve [problem you solve]?'), Comparison queries ('[your category] vs [alternative]'), and Recommendation queries ('who should I hire for [your service]?'). Use 5 queries per type minimum.

2

Run every query on each platform

Test every query on ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews. Use a private/incognito window to avoid personalization. Record results in a spreadsheet with columns: Query | ChatGPT | Perplexity | Gemini | Claude | AI Overviews. Mark Y (mentioned), L (linked citation), N (not mentioned).

3

Calculate SOM per platform

For each platform: count the Y + L responses → divide by total queries → multiply by 100. Example: ChatGPT cited your brand in 6 of 20 queries = 30% SOM on ChatGPT. Calculate separately for mention rate (Y+L) and linked citation rate (L only).

4

Calculate competitive SOM

Run the same queries and record when competitors appear. Calculate: (Your mentions) ÷ (Your mentions + all competitor mentions) × 100. This is your relative share — if you appear 6 times and competitors appear 14 times combined, your competitive SOM is 30%.

5

Score sentiment

For each mention, record: Positive (brand described favorably or as a recommended solution), Neutral (brand mentioned without evaluation), Negative (brand mentioned with caveats or criticism). Sentiment matters because you can increase citation frequency while losing the narrative if the model mentions you with qualifications.

6

Document and repeat monthly

Save your query set and use it every month. Consistency matters more than any single data point — AI model responses change frequently as indexes update. Monthly tracking lets you connect specific GEO actions (content updates, entity signal work, schema changes) to measurable SOM movement.

Platform-by-Platform SOM Metrics

ChatGPT

64.5% market share

CITATION PATTERN

Frequently mentions without linking

SENTIMENT

0.052 positive score

GA4 undercounts ChatGPT influence — most mentions don't generate trackable clicks. Brand association built without clicks.

Perplexity

780M monthly queries

CITATION PATTERN

Most balanced citation-to-mention ratio

SENTIMENT

76.9% positive sentiment

Highest quality citations — transparent source display. Perplexity sentiment 14.8x more positive than ChatGPT for same brands.

Google AI Overviews

30–40% of all search queries

CITATION PATTERN

59.8% of citations to brand domains

SENTIMENT

Highest brand preference rate

Strongest brand citation preference. Only appears on ~33% of queries, but when it does, brand domains dominate.

Gemini

Google ecosystem

CITATION PATTERN

Inherits Google authority signals

SENTIMENT

Google E-E-A-T weighted

Track via Google Search Console signals. Brands with strong Google rankings naturally have higher Gemini SOM.

Claude

Growing — Brave Search powered

CITATION PATTERN

Multi-source verified citations

SENTIMENT

Penalizes promotional content

Highest quality bar — Claude cites fewer sources but those it cites carry strong credibility signals.

The SOM Measurement Framework

Four metrics form the complete SOM picture. Track all four — optimizing for only citation rate while ignoring sentiment creates a misleading scorecard.

1. Citation Rate

(Brand citations ÷ Total queries) × 100

The core SOM number. Percentage of queries where your brand appears.

CADENCE: Monthly minimum. Weekly for optimized pages.

TARGET: 15%+ = strong. 30%+ = dominant.

2. Linked Citation Rate

(Linked citations ÷ Total queries) × 100

Citations that include a hyperlink to your domain. Generates trackable GA4 referral traffic.

CADENCE: Monthly. Cross-reference with GA4 AI referral sessions.

TARGET: Perplexity and AI Overviews have highest link rates.

3. Competitive SOM

(Your mentions ÷ All category mentions) × 100

Your relative share vs competitors. Reveals market position in AI search.

CADENCE: Quarterly. Use the same competitor set each time for consistency.

TARGET: Growing share matters more than absolute percentage.

4. Sentiment Score

% Positive — % Negative mentions

How AI tools frame your brand when they cite you. Positive vs neutral vs qualified/negative.

CADENCE: Monthly. Flag any negative or qualified mentions for content response.

TARGET: 70%+ positive. Negative mentions require immediate content action.

Downstream signals to track alongside SOM

AI referral traffic (GA4)

Acquisition → Traffic acquisition → filter by session source for AI domains

AI referral conversion rate

Compare conversion rates of AI referral sessions vs organic. Expect 9x difference.

Branded search lift

If ChatGPT SOM doubles, branded search in GSC typically lifts 15–20% within 4–6 weeks.

Post-conversion survey

Ask new customers 'where did you first hear about us?' Include ChatGPT/Perplexity as options.

SOM Tracking Tools

ToolCostBest forPlatforms covered
Manual spreadsheetFreeBrands starting GEO — monthly manual auditAll 5 platforms
Otterly.ai~$200–500/moMid-market, citation + competitive benchmarkingChatGPT, Perplexity, AI Overviews
Siftly~$200–400/moCitation rate tracking + sentiment analysisChatGPT, Perplexity, AI Overviews
Superlines~$300–600/moMulti-platform, weekly automated trackingChatGPT, Perplexity, Gemini, AI Overviews
Profound.io$2,000–10,000/moEnterprise — competitive share of model at scaleAll major AI platforms
Cited Free AuditFreeBaseline SOM audit — 50+ queries in 48hrsChatGPT, Perplexity, Gemini, Claude, AI Overviews

Recommendation: Start with manual monthly measurement until SOM reaches 15%+. The workflow is manageable with a spreadsheet at that scale. Move to a paid tool when tracking frequency needs to increase.

How to Improve Your SOM

SOM improvement is driven by GEO work. These are the specific levers ranked by impact on citation rate.

Content freshness

3.2x citation multiplier

Update publication dates and refresh statistics on top pages. Freshness within 30 days is the single fastest citation lever.

Entity signal building

0.664 correlation

Create Wikidata entry, build consistent NAP across directories, earn third-party mentions. Entity recognition is the foundation.

Opening paragraph answers

+115% citation likelihood

Add a direct 1-paragraph answer below every H1. Write as if it will be copy-pasted into a ChatGPT response.

Original data and statistics

+40% visibility

Add original research or proprietary data points to top pages. Become a primary source, not a secondary one.

Bing sitemap submission

Unlocks ChatGPT

Without Bing indexing, ChatGPT cannot cite your pages in web search mode. Submit immediately if not done.

Structured headings

68.7% of cited pages use them

Every H2 should work as a standalone answer. AI extracts chunks — ensure yours work independently.

Brave Search submission

Unlocks Claude

Submit sitemap to brave.com/search/webmaster-tools. This is the most overlooked platform unlock.

Remove promotional tone

-26.19% if present

Audit top pages for sales language. Rewrite promotional sentences as factual, informational prose.

The Closing Window — Why Act Now

The organic AI visibility window is closing. Two developments in 2026 signal that the first-mover advantage for SOM has a hard deadline:

DEVELOPMENT 1

ChatGPT ads are confirmed

OpenAI's CFO has confirmed that ads are coming to ChatGPT. Once paid AI placement exists, organic citation share will be more contested and harder to maintain.

DEVELOPMENT 2

Google AI Mode is testing ads

Google began testing sponsored placements within AI Mode in February 2026. The organic citation layer in AI Overviews will shrink as paid placement expands.

Citation authority compounds over time — the same way domain authority compounded in early SEO. Brands that establish strong SOM in 2026 will have a durable advantage as AI search scales and paid placement becomes the norm.

AI search is projected to surpass traditional search by 2028. The brands building SOM now are the brands AI tools will default to citing in 2027, 2028, and beyond. The window for organic first-mover advantage is open — but it's closing.

Free baseline audit

Get your Share of Model score in 48 hours

We run 50+ queries across ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews and calculate your SOM baseline — with a prioritized list of the 3 highest-impact actions to improve it.

Get your free SOM audit

No call required. Delivered in 48 hours.

Go Deeper: Share of Model Cluster

What is Share of Model (SOM)? The new share of voice

Coming soon

How to track your brand in ChatGPT and Perplexity

Coming soon

AI brand monitoring: tools and methodology

Coming soon

How to benchmark your AI citation rate

Coming soon

Share of Model vs Share of Voice: what's the difference

Coming soon

How to build an AI visibility dashboard

Coming soon

Quarterly GEO reporting: what to track and why

Coming soon

How competitor brands are winning AI search

Coming soon

Related reading

Frequently Asked Questions

What is Share of Model (SOM)?

Share of Model (SOM) is a GEO metric that measures how often your brand is cited in AI-generated answers when users ask questions in your category. It is calculated as: (number of AI responses mentioning your brand) ÷ (total queries tested) × 100. For example, if you test 20 queries on ChatGPT and your brand appears in 4 responses, your SOM is 20% for that platform. SOM is the AI equivalent of Share of Voice in traditional marketing.

How is Share of Model different from Share of Voice?

Share of Voice (SOV) measures your brand's presence in paid and organic search results relative to competitors. Share of Model (SOM) measures your brand's presence in AI-generated answers. The key differences: SOV is measured on a results page; SOM is measured inside generated text. SOV requires a click to reach your brand; SOM embeds your brand directly in the answer. SOV is primarily a Google metric; SOM must be measured across ChatGPT, Perplexity, Gemini, Claude, and AI Overviews separately since only 11% of domains overlap between platforms.

What is a good Share of Model score?

Share of Model benchmarks by category: 0–6% = below average (most B2B brands start here); 7–15% = developing (GEO work is showing early results); 16–30% = strong (recognized category presence in AI search); 31–50% = dominant (category authority — AI tools default to citing you); 50%+ = category ownership (brand is synonymous with the category in AI answers). Most B2B brands starting GEO from zero should expect to reach 15–25% SOM within 90 days of a full GEO program.

How often should I measure Share of Model?

Measure SOM monthly as a baseline cadence. For pages you have recently updated or optimized, re-run your query set every 7 days to detect early citation changes. For competitive benchmarking, measure quarterly. Use automated daily monitoring for high-priority branded queries if budget allows. Manual monthly measurement with a structured query set is sufficient for most B2B brands in the early stages of their GEO program.

Does SOM vary by platform?

Yes — significantly. Only 11% of domains are cited by both ChatGPT and Perplexity, meaning your SOM can be strong on one platform and zero on another. You must measure SOM separately for each platform. Perplexity tends to have the most balanced citation-to-mention ratio. ChatGPT frequently mentions brands without linking. Google AI Overviews showed 59.8% of citations going to brand domains. Sentiment also varies dramatically: Perplexity scores 76.9% positive sentiment vs ChatGPT's 0.052 score — a 14.8x gap.

What tools can I use to track Share of Model?

Tools for tracking SOM range from free manual methods to enterprise platforms. Free: manual query testing in a spreadsheet (20–50 queries per month across each platform). Mid-market: Otterly.ai, Siftly, Superlines — typically $200–$500/month for automated tracking across multiple platforms. Enterprise: Profound.io, Semrush AI Toolkit — $2,000–$10,000/month with competitive benchmarking. For most B2B brands starting GEO, manual monthly measurement is sufficient until SOM reaches 15%+ and the cadence needs to increase.

Can I track AI referral traffic in Google Analytics?

Yes. In GA4, go to Acquisition → Traffic acquisition and filter by Session source. AI referral traffic appears from these domains: chat.openai.com (ChatGPT), perplexity.ai (Perplexity), gemini.google.com (Gemini), claude.ai (Claude). Most B2B brands see AI referral traffic as 3–12% of total traffic in 2026. ChatGPT referral traffic converts at 15.9% vs 1.76% for Google organic — track this as a separate revenue signal, not just a traffic metric.

What is the difference between AI mention and AI citation?

An AI mention is when an AI tool names your brand in its response without linking to your website. An AI citation is when the AI links directly to a page on your domain. Both matter. Mentions build brand association and influence buyer perception without generating trackable clicks. Citations generate measurable referral traffic. Perplexity's citation-to-mention ratio is the most balanced of the major platforms. ChatGPT frequently mentions brands without linking, meaning GA4 undercounts ChatGPT influence significantly.