Share of Model - Why AI visibility is the new marketing KPI

It's happening before our very eyes, and yet many marketers lack awareness of it: the way consumers discover brands is fundamentally shifting. ChatGPT now processes around 2.5 billion prompts a day and is one of the top five most visited websites in the world with 5.7 billion monthly website visits. Perplexity AI answers 780 million queries per month - a growth of 239 percent within one year.

The Comparis data provides an impressive picture for Switzerland: the proportion of consumers using AI tools has almost doubled from 27.4% (2024) to 52.9% (2025). At the same time, traditional search engine use has fallen by around four percent since 2020. In Germany, SE Ranking recorded a growth in AI referral traffic of over 700% with almost 64,000 websites analyzed - still small in absolute terms, but with exponential momentum.

Gartner already predicted at the beginning of 2024 that traditional search volumes would fall by 25% by 2026. This forecast is materializing: so-called zero-click searches, where users receive a direct AI-generated answer and never click on a website, already account for 59.7% of all Google queries in Europe. When Google's AI Mode is active, this figure rises to 93 percent.

What Share of Voice is no longer enough

The concept was coined in 2024 by Jack Smyth, Chief Solutions Officer at Jellyfish (The Brandtech Group). Tom Roach, VP Brand Planning also at Jellyfish, then categorized it in a clear development line in Marketing Week: Share of Market → Share of Voice → Share of Search → Share of Model. The formula is simple: brand mentions by AI models divided by the total mentions in the category, expressed as a percentage.

The decisive difference to Share of Voice lies in the binarity. Search engines also show less popular brands - on page five if necessary. AI models, on the other hand, do not recognize page two. If a brand is not anchored in the knowledge model, it simply does not exist in the AI-generated reality. Researchers at INSEAD business school, in collaboration with Jellyfish, found that many established brands with high consumer awareness are surprisingly weakly represented in AI responses - so-called "high-street heroes" that are strong with humans but invisible with machines.

There is also a considerable variance between models. In one analysis, the detergent brand Ariel showed around 24 percent share of model with Meta's Llama, but less than one percent with Google's Gemini - in the same market. Brands therefore need to be measured across multiple AI platforms.

How AI models decide which brands to recommend

Language models operate via two paths. Firstly, via parametric knowledge learned from large amounts of data during training. Secondly, via real-time retrieval using web search or Retrieval Augmented Generation (RAG). At ChatGPT, an estimated 60 percent of queries are answered purely from training data - without web search.

The decisive factor for parametric knowledge is how frequently and authoritatively a brand is mentioned in the training data. Wikipedia content accounts for around 22 percent of the training data of large language models; for ChatGPT, Wikipedia is the most cited source at 47.9 percent. Contrary to SEO intuition, the strongest statistical predictor of AI visibility is not the number of backlinks, but the brand search volume. Brands that are mentioned on four or more platforms are 2.8 times more likely to appear in ChatGPT responses.

Each AI platform has its own preferences: Perplexity obtains 46.7 percent of its citations from Reddit. Google AI Overviews relies on Reddit for 21 percent and YouTube for 18.8 percent. ChatGPT's web browsing correlates with 87 percent of Bing's top organic results.

Five levers for more AI visibility

The most academically sound study in this field comes from researchers at Princeton University and Georgia Tech. Their "Generative Engine Optimization" study published at KDD 2024 analysed 10,000 search queries and showed that targeted optimization can increase AI visibility by up to 40 percent. Five levers proved to be particularly effective.

Structured data and schema markup: Machine-readable context helps AI systems to interpret content precisely. Experiments show that schema markup improves the response accuracy of ChatGPT by 30 percent. Comparison tables with clean HTML achieve 47 percent higher citation rates. Priority is given to FAQPage, Organization, Product and Article schema in JSON-LD format.

Authoritative, quotable content: The Princeton study shows that citing sources in content can increase visibility by up to 115 percent - quotes by 37 percent, statistics by 22 percent. Content should provide direct answers in the first 150 words and be structured in 40-60 word paragraphs. Pages that have been updated within the last 30 days receive 67 percent more AI citations.

Wikipedia and Knowledge Graph presence: 50 percent of the top marketing agencies cited by large language models have a Wikipedia entry. Wikidata entries with consistent entity information and multilingual Wikipedia pages disproportionately strengthen the basis for AI brand recognition.

Digital PR and brand mentions on highly authoritative sources: Earned media flows directly into training data and retrieval systems. Reddit, YouTube and industry publications are among the most frequently cited sources across all major AI platforms. Nick Taylor from Edelman described earned media as the most important driver of brand visibility in AI-generated responses.

Technical optimization for AI crawlers: The robots.txt must explicitly allow GPTBot, ClaudeBot and PerplexityBot. The new llms.txt standard, already implemented by Stripe, Cloudflare and over 600 other websites, offers AI crawlers a summary of the website optimized for the machine. Server-side rendering is mandatory - AI crawlers do not execute JavaScript.

What CMOs in the DACH region should do now

The first step is an audit. Test 50 to 100 prompts that reflect typical customer queries in your category - in parallel on ChatGPT, Claude, Gemini and Perplexity. Important: Run each query multiple times. Rand Fishkin's research with 600 subjects showed that only 30 percent of brands remain consistently visible between successive AI responses. Make sure you also test in German: "Tell me everything you know about brand [X]" delivers different results than the English version.

Integrate AI visibility metrics into your existing dashboards. Relevant KPIs include mention frequency, citation rate relative to competitors, sentiment score and - crucially - the conversion rate of AI referral traffic. The data is compelling: AI search traffic converts at 14.2 percent, compared to 2.8 percent for traditional Google searches. That is five times higher.

Professional tools are now available for monitoring: Profound (155 million dollars in funding, over ten percent of Fortune 500 companies as customers), Otterly.ai (with Swiss partner AB3.ch for the DACH region) and Semrush with integrated AI visibility tracking. Jellyfish offers its own share-of-model platform.

In terms of budget: AI visibility is an additional investment, not a redistribution of the SEO budget. Deloitte Germany classifies the combined approach of SEO, AEO (Answer Engine Optimization) and GEO (Generative Engine Optimization) as a "strategic necessity". Companies with a high level of GEO maturity invest almost twice as much as their competitors - and 97 percent of digital managers report a positive impact.

The DACH region has a head start - yet

The first-mover advantage in German-speaking countries is real. According to a KfW analysis from February 2026, only 20 percent of DACH SMEs use AI at all - and practically none of them ask themselves whether they appear in AI-generated responses. Some specialized agencies are already positioning themselves: Claneo (Berlin), eMinded (Munich), Dachcom and AB3.ch (Switzerland), the GEO Agency Zurich. But the field is still wide open.

The core message for marketing decision-makers is clear: share of model is not the next hype KPI. It captures whether a brand exists in the AI-mediated reality in which a growing majority of consumers make their decisions. INSEAD research shows that strong traditional brand awareness does not automatically translate into AI visibility. And AI referral traffic is growing by 700 percent annually with five times the conversion rate.

Investing now in structured data, authoritative content, knowledge graph presence, digital PR and technical AI accessibility will build a cumulative advantage. Those who wait risk what a recent white paper from Monks aptly puts: being optimized by the machines from the conversation.

About the Author: Lucas Blochberger is Founder & CEO at Blck Alpaca in Vienna, an agency specializing in data-driven marketing and AI automation. Blck Alpaca develops tailor-made AI agents for marketing automation and supports companies in the DACH region in optimizing their processes - from content creation and lead generation to data-driven campaign management. References include projects with IPEC Group, Heimwatt and Biopower. blckalpaca.at