By maxaeo.ai ｜ Published 2026-10-05 ｜ Updated 2026-10-05
Share of model vs share of search compares two different signals: how much existing demand a brand captures in traditional search and how often AI systems surface that brand during discovery and evaluation. Used together, they reveal whether awareness and AI recommendation strength are moving in the same direction—or creating a hidden growth gap.
What Do Share of Search and Share of Model Measure?
Share of Search measures a brand’s proportion of category-level branded search interest. Share of Model measures its proportion of visibility within a controlled collection of AI-generated answers. The first reflects what buyers actively look for; the second reflects what answer engines place in front of them.
A practical Share of Search formula is:
Brand search interest ÷ Combined search interest for tracked brands × 100
Google Trends is commonly used for relative comparisons. Its data is sampled, normalized by geography and time, and scaled from 0 to 100 rather than reported as absolute query volume. All competitors must therefore be compared under the same settings. (support.google.com)
A basic Share of Model formula is:
Brand mentions ÷ All tracked brand mentions in the same AI answer set × 100
Unlike search demand, this denominator is researcher-defined. Results depend on the prompts, engines, markets, answer runs, competitors, and scoring rules included in the study.
How Are the Two Metrics Different?
Share of Search is an audience-demand metric, while Share of Model is an algorithmic recommendation metric. Neither replaces the other because they observe different stages of discovery and consideration.
| Dimension | Share of Search | Share of Model |
|---|---|---|
| Primary signal | Branded search demand | Presence in AI answers |
| Typical data source | Google Trends or keyword-volume data | Recorded LLM and answer-engine outputs |
| Denominator | Search interest across a competitor set | Visibility across a prompt and competitor set |
| Buyer behavior | User already knows what to search for | User asks for advice, options, or comparisons |
| Main variables | Geography, period, category, spelling | Prompt, engine, model, run, position, citation |
| Best use | Tracking brand awareness and demand | Tracking AI discovery and recommendation |
| Core limitation | Misses journeys completed inside AI interfaces | Highly sensitive to sampling design |
Google notes that AI Overviews and AI Mode can use different models and techniques, producing different answers and supporting links. This makes cross-engine measurement—not one isolated screenshot—essential. (developers.google.com)
How Should Share of Model Be Calculated?
A defensible calculation uses a fixed prompt inventory, repeated collection rules, explicit position weights, and a stable competitor universe. A percentage without those controls is not comparable over time.
Use this five-step method:
- Define the decision space. Group prompts by problem discovery, category research, alternatives, comparison, and purchase validation.
- Fix the scope. Record language, country, engine, account state, device assumptions, and whether web retrieval is enabled.
- Classify each appearance. Separate a passing mention, shortlist inclusion, primary recommendation, citation, and negative reference.
- Apply transparent weights. For example, assign 3 points to a primary recommendation, 2 to a top-three option, 1 to another mention, and 1 additional point when the brand’s domain is cited.
- Divide the brand’s weighted points by all competitor points.
The weights are management choices, not universal standards. Publish them with every report and preserve raw answers so analysts can audit changes. For a deeper implementation, use this cross-engine Share of Model calculation framework and its companion guide to weighted AI visibility scoring.
What Can the Gap Predict?
The gap between the two metrics indicates whether AI recommendations reinforce established demand, lag behind it, or create awareness before branded searches appear. It is a strategic signal, not a guaranteed forecast of revenue.
The following original Demand–Recommendation Grid turns the comparison into an actionable diagnosis:
| Share of Search | Share of Model | Interpretation | Priority |
|---|---|---|---|
| High | High | Established demand reinforced by AI visibility | Defend citations and positioning |
| High | Low | Buyers know the brand, but AI often omits it | Fix entity, evidence, and source gaps |
| Low | High | AI frequently introduces the brand before direct search | Connect recommendations to conversion paths |
| Low | Low | Weak demand and weak algorithmic visibility | Build category authority and awareness |
The most revealing measure is the Recommendation Gap:
Share of Model − Share of Search = Recommendation Gap
A negative gap may expose an AI consideration problem before it is obvious in traffic reports. A positive gap may signal emerging discovery, but teams should validate it against assisted visits, trials, demos, and pipeline rather than assume causation.
What Does a SaaS Example Look Like?
A worked example shows why equal measurement windows matter. The numbers below are illustrative, not customer results, and demonstrate how a SaaS team can interpret the relationship without treating either metric as market share.
Suppose four software brands receive a combined 100 points of normalized branded search interest. Brand A holds 32 points, producing a 32% Share of Search.
The team then runs 30 buyer prompts across four AI engines, creating 120 prompt-engine observations. Weighted scoring produces 200 competitive visibility points. Brand A receives 36 points, resulting in an 18% Share of Model.
Its Recommendation Gap is therefore:
18% − 32% = −14 percentage points
This does not prove that sales will decline. It says AI answers underrepresent the brand relative to demonstrated search demand. The team should inspect missing prompts, competitors recommended instead, unfavorable descriptions, and frequently cited third-party sources. Those diagnostic layers are more useful than the headline percentage alone.
How Do You Build a Dual-Track Visibility Scorecard?
Track both metrics on separate, consistent schedules, then connect them to business outcomes. Do not combine them into one opaque score until stakeholders can see how each component behaves.
A practical scorecard should include:
- Share of Search by market and product category
- Unweighted AI mention rate
- Weighted Share of Model
- Average recommendation position
- Positive, neutral, and negative sentiment
- Brand-owned versus third-party citation share
- Visibility by engine and buyer-journey stage
- Recommendation Gap
- AI-assisted visits, trials, demos, and pipeline
- Month-over-month change with methodology notes
Search demand may be reviewed monthly, while AI visibility benefits from daily collection because outputs and sources can change. Teams can connect these indicators through a full-funnel AI search scorecard or a broader B2B SaaS GEO measurement framework.
MaxAEO supports daily monitoring of mentions, citations, recommendations, sentiment, competitive positioning, and source patterns across eight AI engines. Brands can also generate a free AI visibility diagnostic before establishing a recurring benchmark.
Frequently Asked Questions
Is Share of Model the same as an AI company’s market share?
No. It measures a tracked brand’s visibility within defined AI answers. It does not measure the market share, usage, or revenue of ChatGPT, Gemini, Perplexity, or another model provider.
Can Share of Model replace Share of Search?
No. Share of Model captures AI-mediated recommendations, while Share of Search captures expressed brand demand. Removing either metric leaves part of the buyer journey unmeasured.
How many prompts are needed?
There is no universal minimum. Start with enough prompts to cover major buyer stages, personas, use cases, and comparison scenarios. Expand the inventory when adding engines or markets, and keep a stable core panel for trend analysis.
Why can two platforms report different results?
They may use different prompts, engines, competitors, locations, run frequencies, mention definitions, or position weights. Compare methodologies before comparing percentages.
Which metric should executives see first?
Show both alongside the Recommendation Gap and a downstream outcome such as qualified trials or pipeline. This preserves the distinction between existing demand, AI recommendation strength, and commercial impact.