Signal 09 of 09

How often AI is recommending you to buyers today

29/100Currently limiting your visibilityRefanalytics.com

Why this matters

This is the bottom line: how often are AI assistants actually recommending your business to buyers? When someone asks ChatGPT, Perplexity, Claude, Gemini, or Bing Copilot for a recommendation in your category, do you appear?

There is no page two of a ChatGPT response. If AI does not recommend you, the buyer never sees you. Not reduced traffic. Nothing. Competitors whose sites are easier for AI to understand are recommended in roughly 60-80 of every 100 buyer queries.

Per-model breakdown

Average across 5 AI platforms: 29/100

ChatGPT15

Heavily weights protocol compliance and transactional readiness. ChatGPT Shopping needs UCP and ACP to include you.

Signal weights:UCP (20%) + ACP (20%) + Structured Data (15%) + Reviews (15%) + FAQ (10%) + Agent Manifest (10%) + llms.txt (10%)

Perplexity40

Weights editorial authority and third-party validation. Press coverage and E-E-A-T are the primary drivers.

Signal weights:E-E-A-T (20%) + Press (15%) + Reviews (15%) + Structured Data (15%) + Case Studies (10%) + FAQ (10%) + llms.txt (10%) + Social (5%)

Claude25

Weights entity verification and balanced signal coverage. sameAs schema linkage is critical.

Signal weights:sameAs Entity (15%) + E-E-A-T (15%) + Structured Data (15%) + Reviews (15%) + Case Studies (10%) + UCP (10%) + llms.txt (10%) + FAQ (10%)

Gemini35

Weights Google-friendly signals. Structured data and social authority are dominant factors.

Signal weights:Structured Data (20%) + E-E-A-T (15%) + Reviews (15%) + Social (15%) + FAQ (10%) + sameAs (10%) + Case Studies (10%) + Press (5%)

Bing Copilot30

Weights social presence and press coverage. LinkedIn is particularly influential.

Signal weights:Social (20%) + Press (15%) + Reviews (15%) + Press Distribution (10%) + Structured Data (15%) + E-E-A-T (15%) + FAQ (10%)

Competitor advantage

Brands with ACP-compatible checkout endpoints are completing AI-initiated transactions in your category while visitors to your site must find the purchase button themselves. Every AI agent recommendation flows to your protocol-ready competitors.

Estimated competitor head start: 22 months. Every month you delay, the gap compounds as AI training data solidifies their advantage.

What flips each model from 'skip' to 'recommend'

UCP ManifestRequired for ChatGPT Shopping inclusion

ACP CheckoutRequired for AI-initiated transactions

sameAs entity verificationCritical for Claude and Gemini trust scoring

E-E-A-T signalsWeighted 15-20% by every major model

Review platform coverage0 of 3+ minimum for recommendation confidence

Press coveragePrimary trust signal for Perplexity and Copilot

Case studiesWeighted by Perplexity, Claude, and Gemini

llms.txt10% weight across all 5 models

What to do next

Fix the signals weighted highest by the model you care about most. If your buyers use ChatGPT, prioritize UCP + ACP. If they use Perplexity, prioritize press + E-E-A-T.
Implement UCP manifest. This is 20% of the ChatGPT recommendation score.
Add sameAs schema. This is 15% of the Claude recommendation score.
Build review profiles. Reviews are weighted 15% by every model.

Cart (0)