How AI decides who to recommend.
AI assistants don't rank a list of ten links — they synthesize one answer from the sources they trust. This is the research behind that mechanism: 11 studies measuring how ChatGPT, Claude, Gemini, Grok, and Perplexity retrieve, trust, and cite real businesses, published by the team building the platform that engineers it.
Run your free AI visibility scanAI Visibility
The 1.2% Problem
SOCi found ChatGPT recommends 1.2% of local business locations against 35.9% in Google's local 3-pack. The Houston AI Visibility Index measured why: across 750 prompts and 3,750 responses, machine-readable identity out-predicted backlinks by roughly 3.4x. Includes the Five-Engine Audit — a reproducible field protocol published without restriction.
The Houston AI Visibility Index — Mid-Year Update
A month after Volume I, the national data caught up to the thesis — and confirmed it. AI local-business discovery jumped from 6% to 45% of consumers in a year, ChatGPT surfaces just 1.2% of a category's businesses, and an 8-engine audit found the same trust ordering the Index measured in Houston. A short, sourced update.
The 2026 Houston AI Visibility Index
A 750-prompt, 3,750-response cross-model study scoring 150 anonymized Houston dental, legal, and home-services cohorts on a single 0–100 composite — how often ChatGPT, Claude, Gemini, Grok, and Perplexity name, cite, and rank them. The median business scores 31.
Hallucination Rate by Model: A Local-Business Benchmark
Across 1,000 controlled prompts about real Houston businesses, we measure how often each frontier model fabricates verifiable facts — locations, hours, certifications, ownership.
Machine Trust
Machine Trust Architecture: From SEO to Entity-Level Authority
Why traditional SEO signals collapse inside large language models, and what the new trust layer — JSON-LD, citation density, consistent NAP — actually looks like in retrieval logs.
The Entity Completion Playbook
Step-by-step protocol for building a machine-readable business profile from zero to ship — schema.org Organization, FAQPage, sameAs graph, areaServed canonicalization.
Frequently asked questions
How does AI decide which businesses to recommend?
AI assistants synthesize a recommendation from sources they've learned to trust — business profiles, review platforms, directories, and pages that answer a buyer's question directly — rather than crawling and ranking the live web the way Google does. The reports below measure that mechanism directly.
What is the AI Visibility Index?
A recurring measurement of how often real businesses in a given metro are named when AI assistants are asked who to recommend, re-run on a cadence so the data reflects drift rather than a single snapshot. See the Houston AI Visibility Index below.
Why does this matter if my business already ranks well on Google?
Ranking and being named are different mechanisms with overlapping but not identical inputs — a page-one Google result can coexist with zero AI recommendations. See the local AEO vs. traditional SEO report below.