AI Search Ranking Factors: What Makes ChatGPT Cite Your Content

AI search ranking factors are the criteria that large language models (LLMs) like ChatGPT, Perplexity AI, and Google SGE use to select which sources to cite when generating answers to user queries. Unlike traditional Google ranking which prioritizes backlinks and PageRank, AI search evaluates content based on 12 key factors: FAQ schema implementation (3x citation boost), direct answer placement in first 50-100 words, content freshness under 6 months, topical authority through 10+ related articles, quotable statistics with source attribution, E-E-A-T signals (Experience, Expertise, Authoritativeness, Trustworthiness), structured data quality, internal cross-linking density, external citations from authoritative sources, content depth of 2,300+ words, neutral wiki-voice tone, and real-time accessibility. According to OpenAI's technical documentation and Search Engine Land's 2026 GEO research, these factors combine to create a relevance score that determines if your content appears in the typical 3-8 sources cited per AI-generated response.

The 12 Critical AI Search Ranking Factors

1. FAQ Schema Implementation (HIGHEST IMPACT)

Impact Level: CRITICAL (3x citation rate increase)

What It Is:
Structured data markup using schema.org/FAQPage format that makes question-answer pairs machine-readable for LLMs.

Why It Matters:
Search Engine Land's 2026 research shows pages with FAQ schema get cited 300% more often than pages without it. LLMs can extract and quote FAQ answers directly without parsing full page content, dramatically reducing processing overhead.

How LLMs Use It:

1. User queries "What is GEO?"
2. LLM retrieves pages about GEO optimization
3. Pages with FAQ schema get prioritized in source selection
4. LLM extracts acceptedAnswer text from JSON-LD
5. Validates against visible page content
6. Cites source with direct quote from FAQ

Implementation:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is GEO?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "GEO (Generative Engine Optimization) is the practice of optimizing content to appear as cited sources in AI-generated answers..."
      }
    }
  ]
}
</script>

Success Metrics:

• Minimum 5-8 FAQ questions per article
• 40-80 words per answer (quotable length)
• 100% validation in Google Rich Results Test
• Visible FAQ on page (not hidden in closed accordions)

2. Direct Answer Placement (CRITICAL)

Impact Level: CRITICAL (Core GEO principle)

What It Is:
Providing the answer to your article's title question in the first 50-100 words, before any introduction or context.

Why It Matters:
LLMs use the opening paragraph to determine if a page answers the query. If your answer is buried after 500 words of introduction, the LLM may skip to a competitor with a direct answer.

Traditional SEO Format (WRONG):

# What is AI Search?

In this comprehensive guide, we'll explore the fascinating evolution 
of search technology from keyword matching to artificial intelligence. 
First, let's discuss the history of search engines dating back to the 
1990s when... [500 words before answering the question]

GEO-Optimized Format (CORRECT):

# What is AI Search?

AI search uses large language models like ChatGPT and Perplexity to 
understand user intent and generate direct answers by analyzing multiple 
sources in real-time, rather than just returning ranked lists of web pages. 
Unlike traditional search which matches keywords, AI search synthesizes 
information from 3-8 cited sources to provide conversational responses 
with attribution.

[Rest of article provides depth, examples, and supporting details]

Testing Your Content:

• Can someone understand your answer in first 50 words?
• Does it include 1-2 key statistics or facts?
• Is tone neutral "wiki-voice" (not promotional)?
• Would this paragraph work as a standalone quote?

Citation Impact:
Articles with direct answers in first 100 words get cited 4.2x more often than articles that "bury the lede" according to Perplexity's internal analysis shared at SearchLove 2026 conference.

3. Content Freshness (HIGH IMPACT)

Impact Level: HIGH (LLMs strongly prioritize recency)

What It Is:
How recently content was published or updated, with emphasis on current year data and examples.

Why It Matters:
According to OpenAI's technical documentation, ChatGPT's retrieval system applies a recency boost to sources with:

• Publish/update dates within last 3 months (strong boost)
• Current year statistics (2026 data preferred over 2025)
• Recent event references (mentions of latest developments)

Recency Scoring (OpenAI Research):

How to Maintain Freshness:

Monthly Update Checklist:

• ✅ Update statistics (find latest research/reports)
• ✅ Change year references (2025 → 2026)
• ✅ Add recent examples (new tools, case studies)
• ✅ Expand FAQ with new user questions
• ✅ Refresh screenshots (if UI changed)
• ✅ Update external links (replace dead links)
• ✅ Change publish date in metadata

Version Control Example:

*Last updated: February 7, 2026*
*Previous update: January 2026*

Recent updates:
- Added Gartner 2026 AI search statistics
- Updated ChatGPT citation rate data (SEL Feb 2026)
- Expanded FAQ section with 3 new questions
- Refreshed tool pricing (2026 rates)

4. Topical Authority (HIGH IMPACT)

Impact Level: HIGH (Cluster effect)

What It Is:
Having 10-15 interconnected articles on the same topic cluster, demonstrating comprehensive expertise.

Why It Matters:
LLMs evaluate domain authority by analyzing:

• Total articles on topic (10+ signals expertise)
• Internal link density (well-connected cluster)
• Content depth (comprehensive vs. shallow)
• Topic breadth (covering all sub-topics)

Single Article vs. Topic Cluster:

Scenario A: Orphan Article (LOW authority)

yourdomain.com/chatgpt-seo-guide

Result: Cited 2-4% of the time for "ChatGPT SEO" queries

Scenario B: Topic Cluster (HIGH authority)

Core Topic: ChatGPT SEO (15 articles)

yourdomain.com/chatgpt-seo-guide (pillar - 3,000 words)
yourdomain.com/how-chatgpt-chooses-sources (2,400 words)
yourdomain.com/chatgpt-citation-tracking (2,600 words)
yourdomain.com/chatgpt-vs-traditional-seo (2,200 words)
yourdomain.com/chatgpt-visibility-metrics (2,400 words)
yourdomain.com/faq-schema-implementation (2,800 words)
yourdomain.com/geo-optimization-guide (3,200 words)
... [8 more related articles]

Result: Cited 18-25% of the time for "ChatGPT SEO" queries

Building Authority:

Phase 1 (Weeks 1-4): Foundation

• Publish 10-12 core articles on topic
• Cross-link each article to 3-5 related pieces
• Create "Related Resources" section in all articles

Phase 2 (Weeks 5-8): Expansion

• Add 8-10 supporting articles on sub-topics
• Build hub-and-spoke link structure
• Update older articles with links to new content

Phase 3 (Weeks 9-12): Authority

• Publish 5-8 advanced/case study articles
• Earn external citations from industry sources
• Track citation growth (should compound)

Authority Signal Metrics:

5. Quotable Statistics with Attribution (HIGH IMPACT)

Impact Level: HIGH (Data credibility)

What It Is:
Including specific numbers, percentages, and data points with clear source attribution.

Why It Matters:
LLMs prioritize sources that provide verifiable claims. According to Search Engine Land's roundtable research, pages with 5+ statistics get cited 3x more than pages with vague claims.

Citation-Worthy Statistics Format:

❌ WRONG (Vague):

Studies show GEO works better than traditional SEO.
Most marketers are seeing good results with AI optimization.

✅ CORRECT (Specific + Sourced):

According to Search Engine Land's 2026 GEO Benchmark Study, websites 
implementing comprehensive GEO strategies see 340% higher citation rates 
compared to sites optimized only for traditional SEO. Gartner's 2026 
Digital Marketing Survey found that 73% of marketers now track AI search 
citations as a primary KPI.

Statistic Quality Hierarchy:

1. Tier 1 (Best): Industry research with year + specific number

   • "Gartner 2026 survey: 73% of marketers track AI citations"

2. Tier 2 (Good): Company studies with specific data

   • "BrightEdge research shows 37% YoY decline in organic clicks"

3. Tier 3 (OK): General claims with source

   • "According to OpenAI, ChatGPT prioritizes recent content"

4. Tier 4 (Weak): Unsourced claims

   • "Many experts believe GEO is important"

Authoritative Sources for Statistics:

• Industry Research: Gartner, Forrester, McKinsey, BCG
• SEO Publishers: SearchEngineLand, Moz, SEMrush, Ahrefs
• Marketing Data: HubSpot, Content Marketing Institute
• Tech Analysis: a16z, Benedict Evans, Platformer
• Academic: Google Scholar peer-reviewed papers

How Many Statistics:

• Minimum: 3-5 per article (basic credibility)
• Optimal: 8-12 per article (high authority)
• Maximum: 20+ for data-heavy benchmark reports

6. E-E-A-T Signals (MEDIUM-HIGH IMPACT)

Impact Level: MEDIUM-HIGH (Google SGE especially)

What It Is:
Experience, Expertise, Authoritativeness, Trustworthiness signals that validate content quality.

Why It Matters:
While LLMs don't explicitly score E-E-A-T like Google, they do evaluate similar signals when selecting sources.

E-E-A-T Components:

Experience:

• Author byline with credentials
• "About the Author" with real expertise
• Case studies from actual implementations
• Screenshots/data from real usage

Expertise:

• Author qualifications (certifications, role)
• Company background (About page)
• Industry recognition (awards, speaking)
• Publications on authority sites

Authoritativeness:

• Domain age and history
• External citations (who links to you?)
• Brand recognition in SERPs
• Social proof (mentions, shares)

Trustworthiness:

• HTTPS security
• Contact information visible
• Privacy policy and terms
• Professional design (not spam)
• No excessive ads or pop-ups

Implementation Checklist:

✅ Author byline with credentials
   Example: "By Sarah Chen, Senior SEO Strategist"

✅ Author bio with expertise
   "Sarah has 8 years optimizing for AI search, previously at Google"

✅ Organization schema
   Links domain to real company entity

✅ About page with team info
   Shows real people behind content

✅ External citations
   Link to 2-3 authority sources per article

✅ Updated contact info
   Real email, not just forms

Citation Impact:
Pages with clear authorship and credentials see 23% higher citation rates in ChatGPT and 31% higher in Perplexity per Otterly.ai's analysis of 10,000 citations.

7. Structured Data Quality (MEDIUM-HIGH IMPACT)

Impact Level: MEDIUM-HIGH (Technical foundation)

What It Is:
Clean, validated schema markup beyond just FAQPage (Article, HowTo, Organization).

Why It Matters:
Multiple schema types create richer machine-readable context for LLMs.

Schema Priority Stack:

Must Have (Every Page):

1. FAQPage - If FAQ section present (CRITICAL)
2. Article - All blog posts
3. Organization - Site-wide in header/footer

Should Have (Content Type Specific): 4. HowTo - Step-by-step guides 5. BreadcrumbList - Navigation context 6. WebPage - Generic page markup

Nice to Have (Advanced): 7. Review - Product comparisons 8. VideoObject - Embedded tutorials 9. Person - Author profiles

Implementation Example:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Article",
      "headline": "AI Search Ranking Factors",
      "author": {
        "@type": "Person",
        "name": "Cleversearch Team"
      },
      "publisher": {
        "@type": "Organization",
        "name": "Cleversearch",
        "logo": {
          "@type": "ImageObject",
          "url": "https://cleversearch.com/logo.png"
        }
      },
      "datePublished": "2026-02-07",
      "dateModified": "2026-02-07"
    },
    {
      "@type": "FAQPage",
      "mainEntity": [ /* FAQ items */ ]
    }
  ]
}
</script>

Validation:

• 0 errors in Google Rich Results Test
• Validated in GSC Enhancements tab
• Visible in Schema.org validator

8. Internal Cross-Linking Density (MEDIUM IMPACT)

Impact Level: MEDIUM (Topic graph signals)

What It Is:
Number and quality of internal links between related articles in your topic cluster.

Why It Matters:
LLMs can follow internal links to understand topic relationships and content depth.

Linking Strategy:

Per Article:

• Link to 3-5 related articles from same topic cluster
• Use descriptive anchor text (not "click here")
• Create "Related Resources" section before FAQ
• Link from FAQ answers when relevant

Example "Related Resources" Section:

## Related Resources

**From this series:**
- [FAQ Schema Implementation](/blog/faq-schema-guide) - 
  Technical walkthrough for adding structured data
- [Complete GEO Strategy](/blog/geo-strategy-guide) - 
  Comprehensive AI search optimization framework
- [Track ChatGPT Citations](/blog/track-citations) - 
  Monitor your AI search visibility

**External research:**
- [SearchEngineLand: GEO Guide](https://searchengineland.com) - 
  Industry research on generative search
- [Gartner: Future of Search](https://gartner.com) - 
  Enterprise AI search adoption data

Link Quality Scoring:

Target Metrics:

• 3-5 internal links per article minimum
• 8-12 for pillar/hub articles
• 100% of articles linked from at least 2 other articles (no orphans)

9. External Authority Citations (MEDIUM IMPACT)

Impact Level: MEDIUM (Borrowed authority)

What It Is:
Linking to and citing authoritative external sources in your content.

Why It Matters:
LLMs evaluate if you're part of the authority network by checking who you cite and who cites you.

Citation Strategy:

Per Article, Include 2-3 External Citations:

Tier 1 Sources (Cite These):

• SearchEngineLand, Moz (SEO industry)
• Gartner, Forrester, McKinsey (research)
• Google, OpenAI technical docs (platform docs)
• Academic papers (peer-reviewed research)
• Industry leaders (HubSpot, Semrush data)

Citation Format:

❌ WRONG: "Research shows GEO works."

✅ CORRECT: "According to Search Engine Land's 2026 GEO Benchmark 
Study, pages with FAQ schema see 300% higher citation rates."

[Later in Related Resources]
- [SearchEngineLand: GEO Research](https://searchengineland.com) - 
  Industry-leading analysis of AI search optimization

Impact:
Articles citing 2-3 Tier 1 sources see 18% higher citation rates than articles with no external citations per Cleversearch analysis.

10. Content Depth (MEDIUM IMPACT)

Impact Level: MEDIUM (Comprehensiveness signal)

What It Is:
Total word count and topic comprehensiveness.

Why It Matters:
LLMs can extract more quotable passages from comprehensive content.

Word Count Guidelines:

Depth vs. Fluff:

Content depth ≠ word stuffing. Quality depth includes:

• ✅ Multiple sub-topics covered
• ✅ Examples and case studies
• ✅ Data tables and comparisons
• ✅ Step-by-step walkthroughs
• ✅ FAQ section (5-8 questions)
• ✅ Troubleshooting common issues

Not just:

• ❌ Repetitive paragraphs
• ❌ Keyword stuffing
• ❌ Unnecessary introductions
• ❌ Filler content

Testing Depth:

• Can article be split into 3-5 distinct sections?
• Does it answer 80% of related sub-questions?
• Would an expert find new information here?

11. Neutral Wiki-Voice Tone (MEDIUM IMPACT)

Impact Level: MEDIUM (Objectivity signal)

What It Is:
Writing in educational, non-promotional "Wikipedia-style" tone.

Why It Matters:
LLMs are trained to recognize and prioritize objective, educational content over sales copy.

Tone Comparison:

❌ PROMOTIONAL (Wrong):

Our amazing platform is the best solution for GEO optimization! 
We've helped thousands of clients achieve incredible results. 
Sign up today and transform your AI search visibility!

✅ WIKI-VOICE (Correct):

GEO optimization platforms typically include features such as 
citation tracking, keyword monitoring, and competitive analysis. 
Popular tools include Cleversearch, Otterly, and custom API 
integrations, each offering different feature sets and pricing 
tiers based on business needs.

Wiki-Voice Checklist:

• ✅ Third-person perspective (not "we" or "you" constantly)
• ✅ Balanced comparisons (mention alternatives)
• ✅ Fact-based claims with sources
• ✅ Educational intent (teaching vs. selling)
• ✅ Acknowledges limitations/tradeoffs
• ✅ Cites competitors fairly

When to Use Each Tone:

• Wiki-voice: Main article body, definitions, comparisons
• Conversational: FAQ answers, how-to steps
• Promotional: Call-to-action sections only (end of article)

12. Real-Time Accessibility (MEDIUM IMPACT)

Impact Level: MEDIUM (Technical availability)

What It Is:
LLM's ability to access and parse your content in real-time during retrieval.

Why It Matters:
If LLMs can't access your page, you can't be cited.

Accessibility Checklist:

✅ Fast Loading:

• Largest Contentful Paint < 2.5s
• Server response time < 600ms
• No render-blocking resources

✅ Mobile-Friendly:

• Responsive design
• Readable font sizes
• Touch-friendly navigation

✅ No Paywalls:

• Critical content visible without login
• FAQ/answers not behind paywall
• Core value available to crawlers

✅ robots.txt Allows Crawling:

# Allow AI crawlers
User-agent: GPTBot
Allow: /

User-agent: ChatGPT-User
Allow: /

User-agent: PerplexityBot
Allow: /

✅ Clean HTML:

• Valid HTML5
• No JavaScript-only content (for core answers)
• Semantic markup (h1, h2, h3 hierarchy)

❌ Citation Blockers:

• Aggressive bot blocking
• Heavy JavaScript that fails to render
• Paywall on core content
• 403/404 errors for crawlers
• Excessive ads blocking content

How AI Search Ranking Factors Work Together

The Citation Selection Process

Step 1: Query Understanding

• User asks: "How to optimize for ChatGPT"
• LLM expands to related queries: "ChatGPT SEO", "GEO optimization", "AI search visibility"

Step 2: Retrieval (Top 100 Candidates)

• Web search for expanded queries
• Initial filtering by relevance
• Result: 100 potentially relevant pages

Step 3: First-Pass Ranking (Top 20)

Factors Applied:
✓ Content freshness (recency boost)
✓ FAQ schema presence (3x boost)
✓ Direct answer in first 100 words
✓ Page load speed

Result: 20 high-quality candidates

Step 4: Deep Ranking (Top 8)

Factors Applied:
✓ Topical authority (cluster analysis)
✓ Statistics with sources (credibility)
✓ E-E-A-T signals (trust)
✓ Content depth (comprehensiveness)

Result: 8 final candidates for citation

Step 5: Citation Selection (3-5 Sources)

Factors Applied:
✓ Query relevance score
✓ Quote quality (specific passages)
✓ Source diversity (different domains)
✓ Platform-specific preferences

Result: 3-5 sources cited in answer

Scoring Example

Hypothetical Article: "chatgpt-seo-guide"

Ranking Factor Scorecard:

FAQ Schema: ✅ Present (+300% boost)
Direct Answer: ✅ First 50 words (+4.2x boost)
Freshness: ✅ Updated Feb 2026 (+100% boost)
Topical Authority: ✅ 15-article cluster (+6x boost)
Statistics: ✅ 10 sourced stats (+3x boost)
E-E-A-T: ✅ Author credentials (+23% boost)
Structured Data: ✅ Article + FAQ schema (+moderate)
Internal Links: ✅ 7 cross-links (+moderate)
External Citations: ✅ 3 Tier 1 sources (+18% boost)
Content Depth: ✅ 3,200 words (+high probability)
Wiki-Voice: ✅ Neutral tone (+moderate)
Accessibility: ✅ Fast loading, mobile-friendly (+baseline)

Combined Relevance Score: 94/100
Estimated Citation Probability: 35-45% for target keywords

Optimizing for Maximum Citations

30-Day Quick Wins

Week 1: Foundation (Highest ROI)

• Add FAQ schema to all existing articles (15-30 min per article)
• Rewrite first paragraphs with direct answers
• Update publish dates and refresh 2026 statistics

Week 2: Authority Building

• Plan 3-5 topic clusters (10-15 articles each)
• Create internal linking map
• Add "Related Resources" sections to existing content

Week 3: Content Quality

• Expand articles under 2,000 words to 2,500+
• Add 8-12 statistics with sources to key articles
• Implement Article + FAQPage schema everywhere

Week 4: Technical Optimization

• Validate all schema (0 errors in Rich Results Test)
• Optimize Core Web Vitals
• Add author bios with credentials

Expected Results:

• Week 1: 50-100% citation rate increase (FAQ schema impact)
• Week 2: 20-30% increase (authority signals building)
• Week 3: 15-25% increase (content quality improvements)
• Week 4: 10-15% increase (technical polish)
• Total 30-Day Lift: 95-170% citation rate increase

90-Day Strategic Plan

Phase 1 (Days 1-30): Foundation

• Publish 20-25 foundational articles
• Implement all 12 ranking factors
• Build 2-3 core topic clusters
• Target citation rate: 8-12%

Phase 2 (Days 31-60): Expansion

• Publish 20-25 supporting articles
• Earn external citations (guest posts, research)
• Update Phase 1 content with new links
• Target citation rate: 15-20%

Phase 3 (Days 61-90): Authority

• Publish 15-20 advanced articles
• Launch original research/data
• Expand topic clusters to 15+ articles each
• Target citation rate: 25-35%

Related Resources

From this series:

• Complete GEO Strategy Guide - Comprehensive framework for AI search optimization
• FAQ Schema Implementation - Technical guide to highest-impact ranking factor
• Track ChatGPT Citations - Monitor your AI search performance
• ChatGPT SEO Best Practices - Tactical optimization checklist

External research:

• OpenAI Technical Documentation - How ChatGPT Search works under the hood
• SearchEngineLand: GEO Research - Industry analysis of AI search ranking patterns
• Gartner: Future of Search 2026 - Enterprise research on AI search adoption and trends

Frequently Asked Questions

What are AI search ranking factors?

AI search ranking factors are the criteria that large language models like ChatGPT, Perplexity, and Google SGE use to select which sources to cite when generating answers. The 12 critical factors include FAQ schema (3x citation boost), direct answer placement in first 50-100 words, content freshness under 6 months, topical authority through 10+ related articles, statistics with sources, E-E-A-T signals, structured data quality, internal linking, external citations, 2,300+ word depth, neutral wiki-voice tone, and real-time accessibility.

Which ranking factor has the biggest impact?

FAQ schema implementation has the biggest impact, increasing citation rates by 300% according to Search Engine Land's 2026 research. Pages with properly implemented schema.org/FAQPage markup get cited 3x more often because LLMs can directly extract and quote FAQ answers without parsing full page content. Adding FAQ schema takes only 15-30 minutes per article but delivers the highest ROI of any GEO tactic.

How does ChatGPT decide which sources to cite?

ChatGPT uses a multi-stage process: (1) Expands query to related searches, (2) Retrieves top 100 potentially relevant pages, (3) Applies first-pass ranking based on freshness, FAQ schema, and direct answers to narrow to 20 candidates, (4) Deep ranks using topical authority, statistics, E-E-A-T, and content depth to select 8 finalists, and (5) Chooses 3-5 sources based on query relevance, quote quality, and source diversity. The entire process happens in 2-3 seconds per OpenAI documentation.

Do backlinks matter for AI search?

Backlinks have minimal direct impact on AI search citations unlike traditional Google SEO. Instead, external citations matter—being cited by authoritative sources like Gartner, SearchEngineLand, or academic papers signals credibility to LLMs. Articles citing 2-3 Tier 1 sources see 18% higher citation rates. Focus on earning citations from authority publications rather than building generic backlink profiles for AI search success.

How important is content freshness for citations?

Content freshness is highly important—LLMs strongly prioritize recent content. OpenAI research shows articles 0-3 months old maintain 100% citation probability, but this drops to 80% at 3-6 months, 50% at 6-12 months, 20% at 12-24 months, and just 5% after 24 months. Update articles monthly with fresh statistics, current year examples, and refreshed publish dates to maintain maximum citation potential.

Can I rank in AI search without topic clusters?

Single orphan articles rarely get cited consistently—LLMs evaluate topical authority by analyzing your entire domain. Topic clusters with 10-15 interconnected articles see 6-10x higher citation rates than single articles. A domain with 15 articles on "ChatGPT SEO" gets cited 18-25% of the time versus 2-4% for single-article domains. Build comprehensive topic coverage to demonstrate expertise and authority.

What word count is best for AI citations?

Optimal word count is 2,500-4,000 words for comprehensive guides and 2,300-2,800 for focused tutorials. Articles under 2,000 words rarely get cited due to insufficient depth. However, depth means comprehensive topic coverage with examples, data, and FAQ—not keyword stuffing or filler. Pillar content at 4,000-6,000 words sees highest citation rates (25-35%) but requires exceptional quality and maintenance.

How do I optimize for multiple AI platforms?

Core ranking factors apply across ChatGPT, Perplexity, Google SGE, and Bing Copilot: FAQ schema, direct answers, freshness, topical authority, and statistics work universally. Platform-specific differences: Google SGE weighs Core Web Vitals more heavily, Perplexity favors academic citations, ChatGPT prioritizes conversational tone, and Bing Copilot integrates with Microsoft ecosystem. Focus on universal factors first (80% of optimization), then test platform-specific tweaks.