# Share of Model: how to calculate your share of AI brand mentions

> ChatGPT and Google AI disagree on brand picks 61.9% of the time. Calculate Share of Model across engines, avoid vanity math, and report it to your CEO.

HTML: https://parse.gl/blog/share-of-model-new-kpi · Site index: https://parse.gl/llms.txt · Published: 2026-04-30 · Updated: 2026-11-20

Share of Model is your brand's percentage of all recorded brand mentions inside AI answers across a fixed set of category prompts. Always report the exact prompt count, engines, competitor denominator, and date window right next to the number. Done with discipline, Share of Model gives leadership a clean AI-discovery benchmark; done sloppily on one model or five prompts, it turns into vanity math.

Three realities from AI visibility research explain why Share of Model needs strict sampling rules from day one.

- **Cross-engine brand disagreement:** 61.9%
- **Brands named per Google AI Overview query:** 6.02
- **Brands named on narrower AI engines:** 2.37
- **Brands staying visible across consecutive runs:** ~30%

## What Share of Model measures and where it came from

The math is a simple ratio: take total mentions of your brand across a fixed category prompt set, divide by total brand mentions across those exact same answers, and multiply by 100.

Jack Smyth, Chief Solutions Officer at Jellyfish, coined the term and launched Jellyfish's Share of Model platform on December 4, 2024 with Danone and Chivas Brothers as beta participants ([Adweek](https://www.adweek.com/media/marketers-new-metric-share-of-model-llms/); [The Drum](https://www.thedrum.com/profile/the-digital-voice/article/jellyfish-launches-share-of-modeltm-platform)). Boston Consulting Group then formalized Share of Model in its 2026 agentic marketing brief alongside answer position, citations, sentiment, and referral traffic ([BCG](https://www.bcg.com/publications/2026/agentic-scenarios-every-marketer-must-prepare-for)). Other labels float around (AI Share of Voice, GEO Score, LLM Mention Rate), yet they calculate the same underlying ratio. Pick Share of Model and hold the definition steady for four straight quarters.

## How Share of Model stacks against Share of Voice and Share of Search

Share of Model doesn't replace classic marketing KPIs. Each metric tracks a different stage of buyer behavior: paying for attention, searching on Google, and asking an AI assistant for a shortlist.

The table below lines all three KPIs up side by side.

| Metric          | What it measures                          | Underlying data                        | What it signals                     |
| --------------- | ----------------------------------------- | -------------------------------------- | ----------------------------------- |
| Share of Voice  | Paid media presence versus rivals         | Ad spend, impressions, media mentions  | Brand awareness lift                |
| Share of Search | Branded search demand versus rivals       | Google Trends and search query volume  | Future category market share        |
| Share of Model  | AI-generated brand mentions versus rivals | Sampled AI answers to category prompts | AI-driven shortlist discovery share |

Compare them directly in executive reviews. A company with 18% Share of Voice and 4% Share of Model is funding paid awareness that AI assistants fail to convert into recommendations.

## Calculate Share of Model without rigging the denominator

The formula fits on one line:

**Share of Model = (mentions of your brand ÷ total category brand mentions) × 100**

Two teams running that exact formula on the same company can still disagree by 30 percentage points if they don't lock down the inputs. Write six counting rules down before your first run:

1. **Freeze the prompt set:** lock 50 to 200 category-defining prompts spanning discovery, comparison, use-case, and validation intents. (A tiny 5-to-10-prompt starter sheet works for a quick gut check per [Hallam](https://hallam.agency/blog/share-of-model-a-key-metric-for-ai-powered-search/), while [Symphonic Digital](https://www.symphonicdigital.com/blog/understanding-share-of-model) recommends 250 to 500 prompts for broad multi-product indexes.)
2. **Freeze the competitor roster:** keep the denominator consistent quarter over quarter so gains reflect answer movement, not spreadsheet edits.
3. **Sample multiple engines separately:** track ChatGPT Search, Google AI Overviews or AI Mode, and Perplexity at minimum, adding Claude or Gemini where your buyers use them.
4. **Repeat runs per prompt:** run multiple samples per prompt per engine (or steady weekly schedules) to smooth stochastic variance, setting API temperature to 0 when measuring via direct APIs.
5. **Define entity matching rules:** state upfront how you count parent brands versus sub-brands, prose mentions versus numbered lists, and exact aliases.
6. **Automate past 25 prompts:** spreadsheets work for 5 to 10 prompts, yet manual copy-pasting breaks repeatability the moment executives ask for quarter-over-quarter trendlines.

## Why single-engine Share of Model hides the real story

BrightEdge measured 61.9% brand-recommendation disagreement between ChatGPT and Google AI on identical queries ([BrightEdge](https://www.brightedge.com/resources/weekly-ai-search-insights/chatgpt-vs-google-ai-62-brand-recommendation-disagreement)). Rolling multiple engines straight into one blended average erases the single most useful diagnostic in your data: which engine recommends you and which hands deals to rivals.

Answer density varies sharply too. BrightEdge observed Google AI Overviews naming an average of 6.02 brands per query compared with 2.37 brands on narrower answer surfaces. Parse's 90-day public index from August 23 to November 20, 2026 (502,198 ChatGPT Search answers and 489,554 Google AI Mode answers) reflects those exact structural splits across categories, including huge citation differences like Reddit earning 62,745 citations on Google AI Mode versus 10,262 on ChatGPT Search (6.1× higher). Always compute Share of Model per engine first, then apply buyer-weighted rollups only for the board slide.

Sample weekly for operators, roll up monthly for marketing leadership, and freeze quarterly snapshots for board decks. Because roughly 30% of brands remain visible across consecutive monitoring runs, quarterly-only checks miss shifts until it's too late to react.

## Where Share of Model breaks down (and the three metrics to pair with it)

Share of Model compresses presence cleanly, yet it stays blind to four critical details:

- It treats a skeptical warning the same as a #1 recommendation ([being named isn't the same as being recommended](https://parse.gl/blog/ai-mention-vs-recommendation)).
- It hides whether you landed first or fifth on a crowded shortlist ([AI only names about five brands per answer](https://parse.gl/blog/how-many-brands-ai-names-per-answer)).
- It misses brand-name conflation with similarly named companies.
- It doesn't tell you which websites drove the answer.

Keep Share of Model as the top-line KPI, and always back it up with three companion diagnostics: **recommendation rank**, **answer framing/sentiment**, and **cited source domains** (see [which domains AI models cite most](https://parse.gl/blog/which-domains-ai-models-cite-most) and [how to build an AI visibility prompt set](https://parse.gl/blog/how-to-build-ai-visibility-prompt-set)).

## How to report Share of Model to your CEO in four slides

Executives want clarity, not 40 prompt screenshots. Keep the board package to four slides:

1. **Today's Share of Model by engine:** print the prompt count, competitor denominator, and date window right in the subtitle.
2. **Four-quarter trendline:** anchor Share of Model directly beside your Share of Voice so capital misallocation jumps off the page.
3. **Gap to #1 and #2:** show which specific prompt clusters explain the leader's margin.
4. **Next quarter's fix plan:** list three owner-assigned actions tied to the exact cited domains behind the gap.

For slide templates and executive framing, pair this guide with [how to report AI visibility to your CEO](https://parse.gl/blog/how-to-report-ai-visibility-to-your-ceo) and [what is AI visibility?](https://parse.gl/blog/ai-visibility-explained).

## Run this workflow with your agent

You don't need brittle manual spreadsheets to calculate repeatable Share of Model every week. Connect Claude, ChatGPT, or Cursor to Parse's MCP workspace at `https://agent.parse.gl/mcp` (searching the last 90 days of the Parse Index, reading excerpts, and using the shared work log are free; new accounts start with $5 in trial credit, and fresh weekly checks cost $0.01 per question per engine with no subscription).

Prompt your connected agent:

> "Call `check_brand` on our domain and our five core rivals, run `query_metrics` across ChatGPT Search and Google AI Mode over the last 90 days to calculate Share of Model per engine alongside recommendation share and top cited domains, pull the latest workspace summary with `get_brief`, and lock our 50-prompt benchmark on a weekly cadence with `track`."

Explore the [track shifts over time workflow](https://parse.gl/use-cases/track-shifts-over-time) or connect your assistant on [Parse MCP](https://parse.gl/mcp).

## Sources

- [Marketers Are Tracking a New Metric: Share of Model, Adweek](https://www.adweek.com/media/marketers-new-metric-share-of-model-llms/)
- [Jellyfish Launches Share of Model Platform, The Drum](https://www.thedrum.com/profile/the-digital-voice/article/jellyfish-launches-share-of-modeltm-platform)
- [Share of Model: a key metric for AI-powered search, Hallam](https://hallam.agency/blog/share-of-model-a-key-metric-for-ai-powered-search/)
- [Agentic Scenarios Every Marketer Must Prepare For, BCG](https://www.bcg.com/publications/2026/agentic-scenarios-every-marketer-must-prepare-for)
- [ChatGPT vs Google AI: 61.9% brand recommendation disagreement, BrightEdge](https://www.brightedge.com/resources/weekly-ai-search-insights/chatgpt-vs-google-ai-62-brand-recommendation-disagreement)
- [Understanding Share of Model, Symphonic Digital](https://www.symphonicdigital.com/blog/understanding-share-of-model)
- [Share of Model platform, Jellyfish](https://shareofmodel.ai/)
- [GEO Metrics: AI Search KPIs for Competitive Visibility in 2026, LLM Pulse](https://llmpulse.ai/blog/geo-metrics/)

## For agents

- Any brand's shares and rivals, with 95% ranges: `https://agent.parse.gl/v1/tools/check_brand?subject=%3Cbrand-domain%3E`
- Search the index: `https://parse.gl/api/public/search?q=<words>`

Reuse: Figures and tables on this page are free to republish with attribution: a link to https://parse.gl/blog/share-of-model-new-kpi beside them. A standard followed link (dofollow) is preferred.

- No account: Public figures and up to 5 examples per list, about 120 brands a day per address.
- Free account: $5 in credit, your brand's first check (20 buyer questions) and a free weekly check.
- After your first top-up: Exports and deeper search results. The first top-up starts at $5.
- Live checks: $0.01 per question per engine.
- Full public answers and pages: 25 free per day, then $0.005 each.
- Public answers older than 90 days: $0.001 each the first time your workspace counts or lists one, then yours.
- An agent can save work before signup: `POST https://agent.parse.gl/v1/workspaces`
- Connect an assistant: `https://agent.parse.gl/mcp`

[Prices](https://parse.gl/pricing.md) · [Map of the site](https://parse.gl/llms.txt)
