Methodology

Research foundation

This methodology is based on analysis of AI recommendation patterns across 2,800+ queries between October 2024 and January 2025. Key findings:

• 68% of AI-recommended websites do not rank in top 3 Google results for the same query
• Sites with structured FAQ blocks are cited 3.2x more frequently than those without
• Presence of schema.org markup correlates with citation (r=0.54) but does not guarantee it
• Average recommendation confidence threshold: 72/100 across major AI assistants

These observations informed the five-category evaluation framework described below.

How the evaluation works

DefaultAnswer evaluates five categories that map directly to AI recommendation behavior.

What DefaultAnswer measures

Entity clarity

Can an AI system clearly determine:

• what the site is
• who it is for
• what it offers

This is inferred from visible titles, headings, definitions, and consistent terminology.

If the entity is ambiguous, AI systems hesitate to recommend it.

Answerability

Can an AI extract direct answers to common user questions?

Examples:

• “What is this?”
• “How does it work?”
• “Who is it for?”
• “Why choose this?”

Pages that require inference instead of extraction reduce confidence.

Commercial clarity

If the site offers a product or service, can an AI understand:

• what is being sold
• how it is offered
• what the pricing or plan context is

Hidden, vague, or implied commercial information lowers recommendation certainty.

Trust and legitimacy

Can an AI justify citing the site as a source?

Signals include:

• clear ownership or company context
• contact or accountability information
• accessible pages without blocking errors

Without legitimacy signals, AI systems avoid citation.

Accessibility and retrievability

Can the site be reliably fetched and read?

This includes:

• stable status codes
• readable HTML content
• no access restrictions that prevent retrieval

If content cannot be retrieved consistently, it cannot be recommended.

How scores and findings are produced

Each category is evaluated using explicit, repeatable checks.

Findings are mapped to:

• observable on-page signals
• concrete gaps
• specific recommendations

DefaultAnswer does not use black-box scores or probabilistic guesses.

Signal taxonomy

DefaultAnswer evaluates 47 discrete signals across the five categories. Each signal is binary (present/absent) or graduated (weak/moderate/strong).

Entity clarity signals (12)

1. Page title contains category identifier
2. H1 heading defines what site offers
3. First-paragraph definition of core offering
4. "What is [entity]" answered within first 500 words
5. Consistent terminology (entity name used 3+ times)
6. Category placement visible ("CRM for small business")
7. Target audience explicitly stated
8. Use case or problem statement present
9. Differentiation from alternatives mentioned
10. Homepage title matches about page description
11. No contradictory category signals
12. Entity type identifiable (product/service/content/tool)

Answerability signals (10)

1. FAQ section present
2. "How it works" section with steps
3. "Who is this for" explicitly answered
4. Pricing information visible or linked
5. Feature list with plain-language descriptions
6. Use case examples provided
7. Comparison content (vs alternatives)
8. Problem/solution structure visible
9. Questions answered in headings
10. Direct quotes extractable without interpretation

Commercial clarity signals (8)

1. Product/service offering explicitly named
2. Pricing page accessible
3. Plan tiers or options visible
4. Free trial or demo mentioned
5. Purchase or signup path clear
6. B2B vs B2C context identifiable
7. Commercial model stated (SaaS/marketplace/service)
8. Value proposition in plain language

Trust and legitimacy signals (9)

1. About page exists and is linked
2. Contact information visible
3. Company or creator name stated
4. Physical location or incorporation mentioned (if applicable)
5. Privacy policy linked
6. Terms of service linked
7. schema.org organization markup present
8. Social proof elements (logos, testimonials, metrics)
9. Consistent branding across pages

Accessibility signals (8)

1. Page returns 200 status code
2. No authentication wall for public content
3. No paywall blocking descriptive content
4. Mobile-responsive viewport
5. No aggressive bot detection on public pages
6. Structured HTML (not SPA rendering issues)
7. No broken internal links in navigation
8. Meta description present and relevant

Signals are weighted by frequency of appearance in successfully cited sources, not by assumed importance.

What DefaultAnswer does not do

DefaultAnswer does not:

• optimize for search rankings
• evaluate backlinks or authority metrics
• simulate AI model internals
• guarantee recommendations

It explains why confidence exists or is limited, based on what is visible and citable.

Why this approach works

AI assistants generate answers, not rankings.

They prefer sources that are:

• clear
• unambiguous
• easy to quote
• safe to justify

DefaultAnswer measures whether your site meets those conditions.