Optiview AEO + GEO Methodology

By Kevin McGovern · Last updated January 16, 2025

Optiview provides a complete AEO + GEO readiness assessment combining structural audits (21 weighted checks) with live citation testing across ChatGPT, Claude, Perplexity, and Brave Search.

Overview

Our dual-layer assessment evaluates both structural readiness and real-world citation performance:

• Structural Audit (21 checks): How well your site is optimized for training bots and answer engines
• Live Citation Testing: Whether your brand actually appears in LLM responses today
• GEO Adjusted Score: Combined metric that rewards both structure and real-world visibility

What We Measure

• AEO (Answer Engine Optimization): 11 checks for answer boxes, featured snippets, and AI Overviews
• GEO (Generative Engine Optimization): 10 checks for LLM citation readiness
• Live LLM Citations: Real queries against ChatGPT, Claude, Perplexity, and Brave
• Training Bot Access: How GPTBot, ClaudeBot, and PerplexityBot view your site

Scoring Framework

Each check is scored on a 0-3 scale:

AEO Checks (11 total)

GEO Checks (10 total)

Render Visibility Penalty

Sites with low render visibility (content only visible after JavaScript execution) receive penalties:

• AEO: -5 points if <30% of content in static HTML
• GEO: -5 to -10 points if <50% of content in static HTML

This reflects the reality that many AI crawlers (GPTBot, ClaudeBot, PerplexityBot) do not execute JavaScript.

Live Citation Testing

Beyond structural checks, we test how your brand appears in live LLM responses:

Query Generation

We use a context-aware prompt system to generate realistic branded and non-branded queries:

• Industry Detection: Classify your site's vertical using weighted signals, JSON-LD, and navigation taxonomy
• Branded Queries: 10 queries using your brand name and common aliases/nicknames
• Non-Branded Queries: 18 queries about your industry, products, and services (without brand name)
• Quality Gates: All queries are validated for brand leakage, relevance, and realism before testing

LLM Sources Tested

• ChatGPT (OpenAI): GPT-4 with web browsing enabled
• Claude (Anthropic): Claude 3 with web search
• Perplexity: Real-time web search and citation
• Brave Search: AI-powered search results

Citation Metrics

For each source, we track:

• Citation rate (% of queries where your domain appears)
• Branded vs non-branded performance
• Citation position and context
• Competitive gaps (queries where competitors appear but you don't)

GEO Adjusted Score

The GEO Adjusted Score combines structural readiness with real-world citation performance:

• Base GEO Score (70% weight): Your structural score from the 10 GEO checks
• Citation Performance (30% weight): Weighted average of citation rates across ChatGPT, Claude, and Perplexity
• Formula: geo_adjusted = geo_raw × 0.7 + citation_bonus × 0.3

This rewards sites that are both structurally sound and actually appearing in LLM responses today.

Bot Identity & Crawling

Our audit system respects website owners and follows best practices:

• User-Agent: OptiviewAuditBot/1.0 (+https://api.optiview.ai/bot)
• Robots.txt: We parse and respect all Allow/Disallow rules
• Crawl Delay: We honor Crawl-delay directives and implement exponential backoff
• Rate Limiting: Configurable delays between requests to avoid overloading servers
• Meta Robots: We respect noindex, nofollow, and noai tags

Full documentation: OptiviewAuditBot Documentation

Research Foundations

Our methodology is based on:

1. Google's public guidance on AI Overviews
2. Academic research on Generative Engines (arXiv:2404.16366)
3. GPTBot user-agent documentation
4. Cloudflare AI crawler documentation
5. Live testing of thousands of queries across ChatGPT, Claude, Perplexity, and Brave Search
6. Analysis of citation patterns across 18+ industry verticals

Audit Architecture

Our system combines multiple techniques for comprehensive assessment:

• Sitemap Discovery: Automatic detection from robots.txt and common locations
• Breadth-First Crawl: Intelligent link extraction prioritizing top-level navigation
• Dual-Mode Rendering: Static HTML + JavaScript rendering for SPA detection
• Schema Validation: JSON-LD parsing and validation against schema.org
• Industry Classification: Hybrid rule-based + AI embedding classifier (18+ verticals)
• Prompt Generation: LLM-native query generation with quality gates and fallback to industry templates

Test Corpus

We maintain a test corpus of reference pages to validate our scoring: aeo-geo-test-v1.csv

Privacy & Data Handling

We take data privacy seriously:

• We only analyze publicly accessible pages
• Audit data is tied to user accounts via magic link authentication
• We do not share audit results or citations data with third parties
• See our Privacy Policy for details

Content License

This methodology and our scoring guide are published under our content license for transparent reuse and citation.

Sources

For a complete list of references and citations, see our Sources Hub.