How AI Decides What to Recommend.
And how GOSH AI measures, tracks, and improves it, using real data from real AI models across every major platform. The foundational reference for the ARDI™ Intelligence Platform.
(UACR™)
tested
observed
ARDI™ measures how AI models recommend brands, and gives you the intelligence to change the outcome.
It tracks how models like ChatGPT, Gemini, Claude, Perplexity, Grok, and Copilot decide who shows up in generated answers and who doesn’t. When someone asks AI for a recommendation in your category, ARDI™ tells you whether your brand is part of that answer, how it’s positioned, and what your competitors are doing differently.
ARDI™ is the discipline of AI Recommendation and Discovery Intelligence, it unifies Generative Engine Optimization (GEO), entity structuring, content authority, and citation amplification under a single framework that measures the one thing that matters: whether AI recommends you.
Explore the Platform- AI Recommendation & Discovery Intelligence
- The UACR™ decision framework
- Generative Engine Optimization (GEO)
- Entity foundation & structured data
- Content authority building
- Citation amplification
- VSP scoring, Volume, Sentiment, Positioning
- Recommendation Share & displacement
- The Think → Discovery → Decision journey
How AI Actually Decides:
The UACR™ Model.
AI doesn’t rank pages. It makes decisions. Every recommendation an AI model gives passes through four stages, and a brand has to clear all four to be named in the answer. This is the spine of the ARDI™ framework.
AI must resolve your brand as a distinct entity, not a generic term, not a competitor, not a hallucinated version of your business. Consistent naming and structured data make you legible.
AI decides whether you’re relevant to what was actually asked. You have to be associated with the right category, attributes, and intent across the Think, Discovery, and Decision journey, or AI surfaces competitors who are.
Models build confidence from the sources they absorb, content, reviews, publications, structured references. Fragmented content or thin third-party validation means AI has nothing authoritative to lean on, so it stays quiet.
The outcome: your brand is spoken aloud in the generated answer, how often (Volume), how positively (Sentiment), and how prominently (Positioning). This is what the ARDI™ Authority Score measures.
The ARDI™ Authority Score.
A composite from 0 to 100 that reflects how strongly AI models recognize, trust, and recommend your brand, calculated from real testing across six AI models, not estimates or projections.
Critical means AI rarely mentions you. Category Leader means you’re AI’s go-to in your category. Every tier tells you exactly what it takes to move up.
SEO ranks pages. AI chooses brands.
Not the same system, and success in one does not guarantee success in the other.
- Returns a list of links
- Ranks on crawlable content & backlinks
- User clicks through to discover
- Measured by page position (1-10)
- “Page two” still exists
- Returns a direct recommendation
- Selects on trust, authority & training signals
- AI synthesizes and chooses for the user
- Measured by inclusion, sentiment & position
- No “page two”, you’re in or you’re out
Do they work together? Yes, strong SEO can feed the authority signals AI relies on. But SEO alone does not guarantee AI visibility. GOSH AI has documented brands ranking on Google’s first page that appear in zero AI recommendations in their category. Invest in both, and you’re positioned for discovery wherever buyers look.
Where does GEO fit? Generative Engine Optimization is one tactical discipline inside ARDI™, alongside entity structuring, content authority, and citation amplification.
The ARDI™ Testing Methodology.
GOSH AI doesn’t estimate visibility, we test it. Real AI-model outputs become scores, signals, and an action plan, run twice a month so you see movement, not a snapshot.
Structured Prompt Testing
Real questions actual buyers ask, run across all six models and organized by Think, Discovery, and Decision intent.
Brand Mention Extraction
Every response recorded; every mention extracted, normalized, and attributed to the right entity, filtering noise, hallucinations, and generic references.
VSP Scoring Per Model
Volume, Sentiment, and Positioning scored 0-100 for each AI platform individually, so you see exactly where each model stands and where the gaps are.
Recommendation Share Analysis
AI answer-share modeling: who owns the conversation, what each competitor captures, and where you can displace them. Surgical, not generic benchmarking.
Misattribution & Hallucination Audit
What AI gets wrong about you, brand confusion, speculative facts, missing product knowledge, with per-model severity ratings and specific corrections.
Authority Roadmap
A phased plan, Entity Foundation, Content Authority, Citation Amplification, mapped to the models and journey stages with the highest-impact opportunity to move.
AI Recommendations Now Shape Every Category.
If customers in your category are asking AI for recommendations, and they increasingly are, your AI visibility decides whether you’re in the answer. GOSH AI tracks over 70 categories across North America, Latin America, Europe, and Oceania.
A voice response gives one recommendation. There is no second place. As conversational AI embeds in cars, speakers, phones, and wearables, ARDI™ measures how brands surface in voice-prompted answers, and what makes yours the one spoken aloud.
Common Questions.
Hundreds of millions of people now use AI models as their primary way to research products, services, and brands. Instead of scanning ten blue links, they ask a question and receive a direct answer with specific brand recommendations. If your brand isn’t part of that answer, you don’t get a second chance, there is no “page two” in AI. The brands that measure their AI visibility today are the ones recommended tomorrow.
UACR™ is the GOSH AI model for how AI decides what to recommend, in four stages: Understanding (AI resolves your brand as a distinct entity), Association (AI connects you to the right category and intent), Citation (AI has authoritative sources that validate you), and Recommendation (AI names you in the generated answer). A brand has to clear all four to be recommended, and the ARDI™ platform measures each stage.
Yes, and it happens more often than most brands realize. Google rankings are based on crawlable content, backlinks, and technical SEO. AI recommendations are based on entirely different factors, how the model was trained, what patterns it absorbed, and whether your authority signals are strong enough to surface in a synthesized answer. GOSH AI has documented cases where brands on Google’s first page do not appear in a single AI-generated recommendation in their category.
GEO is the practice of optimizing content and brand signals so AI systems can detect, trust, and recommend your brand inside generated answers, summaries, and voice responses. It’s one tactical discipline within the broader ARDI™ framework. Where SEO optimizes for search rankings, GEO optimizes for inclusion in AI-generated recommendations, and ARDI™ measures the result through VSP scoring.
GOSH AI is not a marketing agency. It’s an AI Recommendation & Discovery Intelligence company built to measure and improve how AI models recommend brands. The ARDI™ platform tests real prompts across six leading AI models and seven AI surfaces, produces real scores, and tracks real movement over time. Every insight is based on observed AI behavior, not assumptions, not keyword proxies, not estimated impressions.
Most brands begin to see measurable changes in AI visibility within 60 to 90 days, depending on existing authority signals, category competitiveness, and engagement scope. GOSH AI tracks progress through the ARDI™ Authority Score, so movement is visible from the start. Long-term positioning as a Category Leader requires sustained effort across two to three quarters. The fastest way to see where you stand today is a free ARDI™ QuickScore.