metehan.ai Research & SOPs

Appearance

Platform Comparison ​

Fresh

This content reflects Metehan Alp's research findings as of 2025-2026.

Side-by-side comparison of how major AI search platforms handle ranking, citation, freshness, content format, entities, personalization, and citation display.

Feature Comparison ​

FeatureChatGPTPerplexityGoogle AI ModeGoogle Discover
Ranking MethodRRF fusion + Skysight rerankerXGBoost L3 reranker + authoritative domain lists4-stage pipeline (retrieve, rerank, generate, cite) + JetstreamNAIADES recommendation engine + 13 topic clusters
Top Citation SourceAuthoritative domains (Wikipedia, major publishers, established brands)Curated authoritative domain lists per topic categoryPages already ranking in Google's top 10-20Content matching user interest graph + high engagement velocity
Freshness WeightModerate (time-decay via freshness scoring profile)High (time_decay_rate actively penalizes older content)Moderate-High (freshness is 1 of 7 signals, weighted by query type)Very High (new content heavily favored, tombstoning removes stale content)
Content FormatLong-form, structured headings, FAQ format performs bestFactual density, tables, lists preferred. Claims-per-paragraph matters.500-token sections, answer-first structure, clear H1-H2-H3 hierarchyVisual-first (1200px+ images required), mobile-optimized, short-form works
Entity HandlingEntity recognition for topic association; brand-topic linkage strengthens over timeEntity matching against topic categories for domain authority scoringEntity alignment is 1 of 7 ranking signals; named entity overlap between query and contentEntity-based topic clustering for user interest matching
PersonalizationMinimal in search mode; conversation history affects follow-up queriesModerate (subscribed_topic_multiplier boosts domains in user interest areas)Moderate (inherits Google's personalization signals from Search)Very High (entirely driven by user browsing history and interest graph)
Citation StyleInline text citations with source links; PSL-based domain attributionNumbered inline citations with source preview cards; 4-8 sources typicalInline citations within generated response; 3-5 sources typicalNo citations (recommendation cards with title, image, source name)

Optimization Priority by Platform ​

If Your Primary Target is ChatGPT ​

  1. Build domain authority (ChatGPT favors established, authoritative domains)
  2. Structure content with clear headings and FAQ sections
  3. Ensure content is crawlable by GPTBot and OAI-SearchBot
  4. Target breadth across sub-queries (RRF rewards multi-query presence)
  5. Update content regularly (freshness scoring is active)

If Your Primary Target is Perplexity ​

  1. Maximize factual density (more verifiable claims per paragraph)
  2. Use tables and structured lists for data presentation
  3. Build topic authority in your category (authoritative domain lists)
  4. Keep content fresh (aggressive time decay)
  5. Ensure high semantic similarity between your content and target queries

If Your Primary Target is Google AI Mode ​

  1. Structure content in 500-token sections with answer-first format
  2. Rank on Google first (AI Mode pulls from Google's index)
  3. Use comprehensive heading hierarchies (H1 > H2 > H3)
  4. Build entity alignment (use the same entities the query contains)
  5. Optimize for all 7 ranking signals simultaneously

If Your Primary Target is Google Discover ​

  1. Invest in high-quality images (1200px+ width, visually compelling)
  2. Publish frequently (freshness is the dominant signal)
  3. Build topical consistency (NAIADES matches content to user interest clusters)
  4. Optimize for mobile experience
  5. Drive early engagement (engagement velocity affects promotion)

Key Differences That Matter ​

RRF vs. XGBoost vs. 4-Stage Pipeline ​

  • ChatGPT's RRF rewards breadth: ranking for many sub-queries matters more than ranking #1 for one
  • Perplexity's XGBoost rewards authority: being on the authoritative domain list for your topic category gives a structural advantage
  • Google AI Mode's pipeline rewards existing Google rankings: if you do not rank in Google's top 20, you are unlikely to appear in AI Mode

Citation Density ​

  • Perplexity cites the most sources per response (4-8 typical)
  • Google AI Mode cites 3-5 sources
  • ChatGPT varies significantly by query type (1-6 sources)
  • More citations per response means more opportunities to be included, but also more competition per response

Freshness Sensitivity ​

  • Google Discover is most sensitive (stale content gets tombstoned)
  • Perplexity is highly sensitive (active time decay)
  • Google AI Mode is moderately sensitive (freshness weighted by query type)
  • ChatGPT is least sensitive (freshness matters but authority and relevance dominate)

Cross-Platform Strategy ​

For maximum AI visibility across all platforms:

  1. Build on Google first -- Google rankings are a prerequisite for Google AI Mode and a strong signal for ChatGPT
  2. Structure for 500-token chunks -- Aligns with Google AI Mode and helps all platforms extract content cleanly
  3. Maintain freshness -- Update content quarterly at minimum; monthly for competitive topics
  4. Build real authority -- Domain authority, author credentials, and E-E-A-T signals matter on every platform
  5. Use CiteMET distribution -- AI share buttons create direct citation events across all platforms

Further Reading ​