Research Methodology

Overview

AEOfix conducts ongoing research into Answer Engine Optimization (AEO) performance by analyzing client websites, industry data, and AI engine citation patterns. This page explains our research methodology, data sources, and limitations.

Data Collection Methods

1. Client Website Analysis

Sample Size: 500+ client websites across various industries (B2B SaaS, e-commerce, professional services, healthcare, finance)

Time Period: October 2024 - January 2025 (ongoing)

Data Points Collected:

• Website domain and industry category
• Schema.org implementation status (types, validation scores)
• Content structure (H1-H6 hierarchy, FAQ format, word count)
• E-E-A-T signals (author credentials, citations, publication dates)
• Technical performance (Core Web Vitals, mobile optimization)

2. AI Engine Citation Tracking

Methodology: Manual query testing across ChatGPT (OpenAI), Claude (Anthropic), Google Gemini, and Perplexity AI

Query Testing Protocol:

• Create 50-150 industry-specific questions per client based on target keywords
• Test each query across all four AI engines monthly
• Record which domains appear in bibliographies/citations
• Track citation position (1st, 2nd, 3rd among sources)
• Assess answer accuracy (does AI correctly represent source content)

Limitations: Manual testing limits sample size. AI engine responses can vary based on conversation context, user account history, and model version. Our testing represents snapshots at specific points in time.

3. Server Log Analysis

Data Source: Client website server logs (with permission)

Tracking:

• AI crawler activity (user agent identification)
• Crawl frequency and pages accessed
• Referral traffic from AI platforms

Limitations: Not all AI engines identify themselves with distinct user agents. Direct traffic spikes may correlate with AI citations but can't be definitively attributed.

4. A/B Testing (Before/After Analysis)

Methodology: Compare citation rates before and after AEO implementation

Baseline Period: 30 days pre-implementation

Post-Implementation Tracking: 90 days with monthly checkpoints

Control Factors: We attempt to isolate AEO-specific changes, but other variables (seasonality, algorithm updates, competitor activity) may influence results.

Performance Metrics Calculation

Citation Probability Increase

Formula: (Post-Implementation Citation Rate - Baseline Citation Rate) / Baseline Citation Rate × 100

Example Calculation:

• Baseline: 5 citations per 100 queries (5% citation rate)
• Post-Implementation: 22 citations per 100 queries (22% citation rate)
• Increase: (22% - 5%) / 5% × 100 = 340% increase

Reported Statistics: When we state "340% citation increase," this represents the median improvement observed across our client sample. Individual results range from 150% to 600% depending on industry competitiveness and implementation quality.

Schema Type Impact

Methodology: Compare citation rates for pages with specific schema types (FAQPage, HowTo, Article) vs. pages without schema

Cohort Matching: We attempt to match pages by topic, word count, and domain authority to isolate schema impact

Limitations: Correlation does not prove causation. Pages with schema may also have better content structure, making it difficult to attribute improvements solely to schema implementation.

Third-Party Data Sources

Where possible, we cite external research and official statistics rather than relying solely on our internal data:

• OpenAI official announcements - User statistics and product updates
• Ahrefs SEO research - Industry benchmarks for SEO timelines and ranking difficulty
• Similarweb / Sensor Tower - Traffic analytics for AI platforms
• Industry surveys - User behavior and adoption trends

All third-party sources are hyperlinked throughout our content for verification.

Limitations and Disclaimers

What Our Research Cannot Show

• Causation: We observe correlations between AEO optimization and citation increases, but cannot definitively prove causation due to multiple variables
• Universal Results: Performance varies significantly by industry, competition level, and existing domain authority
• Algorithm Details: AI engine ranking algorithms are proprietary and not publicly documented. Our estimates of ranking factor weights (e.g., "35% schema, 30% E-E-A-T") are based on observed patterns, not official confirmation
• Future Performance: AI engine algorithms evolve constantly. Past performance does not guarantee future results

Market Share Estimates

Individual AI engine market share figures (e.g., "ChatGPT has 60% market share") are estimated based on:

• Third-party traffic analytics from Similarweb and Sensor Tower
• Limited official user statistics from company announcements
• Industry analyst reports

These estimates may not reflect actual usage accurately, as companies do not publicly disclose detailed metrics.

Continuous Improvement

We continuously refine our methodology as we learn more about AI engine behavior:

• Expanding client sample size (targeting 1,000+ websites by Q2 2025)
• Developing automated citation tracking tools to replace manual testing
• Partnering with academic institutions for peer-reviewed research
• Conducting controlled experiments with isolated variables

Last Updated: January 9, 2025

Questions About Our Methodology?

If you have questions about our research approach or would like to discuss specific findings, contact us at research@aeofix.com.