Domain Authority is dead. Not because Moz said so, but because the metric was built for a world where humans clicked through ten blue links. That world is gone. In its place, a new measurement has emerged: Share of Model, the probability that an AI recommends your brand when a user asks a relevant question.
What Is Share of Model?
Share of Model (SoM) is the percentage of times your brand appears in AI-generated responses across a set of representative queries for your category. Unlike Domain Authority, which estimates your likelihood of ranking on Google, Share of Model directly measures whether AI systems like ChatGPT, Perplexity, Gemini, and Claude mention you when potential customers ask questions you should be answering.
Think of it this way: Domain Authority was a proxy. It predicted visibility. Share of Model IS visibility. There is no prediction involved. You either appear in the AI response or you do not.
The metric works by running a statistically significant number of queries across multiple AI models and measuring how often each brand gets cited or recommended. The result is expressed as a percentage. If your brand appears in 34 out of 100 relevant AI responses, your Share of Model is 34%.
Why Domain Authority Stopped Matter
Domain Authority was introduced in the late 2000s as a way to estimate how well a page would rank on Google. It worked because Google algorithm relied heavily on link equity, and DA was a reasonable proxy for link strength. For over a decade, it was the north star metric for SEO teams worldwide.
Three things broke that model:
AI search overtook traditional search behavior. ChatGPT alone processes roughly 64.5% of all AI search sessions as of March 2026, according to Stackmatix market share data. Combined with Gemini at 21.5%, AI search now handles 45 billion sessions monthly. None of those sessions use link equity to determine what to show.
Zero-click became the default. Bain research found that 80% of consumers rely on zero-click AI results at least 40% of the time. Seer Interactive analyzed 25.1 million Google AI Mode impressions and found 93% of queries end without a click. If nobody clicks, your DA score does not matter.
AI citation logic is fundamentally different from ranking logic. Google ranks pages based on backlinks, relevance signals, and user behavior. AI models generate recommendations based on entity recognition, structured data availability, content extraction quality, and training data exposure. A site with DA 85 can be invisible to ChatGPT if its content is not structured for AI extraction.
The 2026 AI Search Visibility Report from Omniscient Digital, which analyzed over 23,000 LLM citations, found that 92% of brands are completely invisible in AI search. Many of those brands have strong Domain Authority scores. They rank well on Google. They just do not exist in AI recommendations.
How Share of Model Is Calculated
The calculation is straightforward, but the methodology matters.
Step 1: Define your query set. Select 50 to 200 queries that represent how real users search for your category. These should be natural language questions, not keyword-stuffed phrases.
Step 2: Run queries across models. Submit each query to ChatGPT, Perplexity, Gemini, and Claude. Record every brand mentioned in the response, including the position and context.
Step 3: Calculate per-model share. For each model, divide the number of queries where your brand appears by the total queries.
Step 4: Weight by market share. Multiply each model SoM by that model market share to get a weighted overall score.
Step 5: Track changes over time. Run this weekly or monthly. A 5-point swing in Share of Model over 30 days is significant.
The Signals That Drive Share of Model
Based on data from the Omniscient Digital study and searchless.ai analysis of AI citation patterns, three signals dominate:
1. Entity Authority (Mentions Across 6+ Domains)
AI models learn about your brand from the broader web. If your brand is mentioned on multiple independent domains in relevant contexts, AI models associate your brand with that category. The threshold appears to be around 6 to 8 independent domains.
2. Answer-First Content Structure
AI models extract the first one to two sentences of a response 73% of the time. If your content buries the answer three paragraphs deep, the AI will find a source that leads with it.
3. Structured Data and llms.txt
ChatGPT reads JSON-LD schema. Perplexity parses FAQ structured data. llms.txt provides a machine-readable map of your content. Brands with all three have measurably higher Share of Model scores.
What to Do This Week
- Run a baseline measurement. Use the manual method or a free tool to find out where you stand.
- Implement llms.txt. It takes less than 30 minutes. Less than 5% of websites have one.
- Restructure your top 10 pages for answer-first content. Rewrite the opening sentence of each page to directly answer the primary question.
These three steps alone can shift your Share of Model by 10 to 20 points within a month.
FAQ
What is Share of Model? The percentage of AI-generated responses that mention or recommend your brand across a representative set of category queries.
How is it different from Domain Authority? DA estimates Google ranking potential based on backlinks. SoM directly measures whether AI models recommend your brand.
Why should I care? 900 million people use AI search weekly, and 92% of brands are invisible in AI recommendations.
What drives higher SoM? Entity authority, answer-first content structure, and technical readiness (llms.txt, FAQ schema, clean HTML).
Find out where you stand. Get your free AI Visibility Score in 60 seconds at audit.searchless.ai.
