ChatGPT, Claude, and Perplexity Don't Agree on a Single Source. Your GEO Strategy Shouldn't Either.

The citation gap is wider than anyone thought.

New research published July 20 analyzed 150,000 queries across ChatGPT, Claude, and Perplexity to map source citation patterns. The finding: these platforms don't agree on sources. At all.

Wikipedia — the bedrock of traditional SEO wisdom for featured snippets and knowledge panels — shows up in 34% of ChatGPT citations, 0.1% of Claude citations, and 0% of Perplexity citations.

If your GEO strategy assumes a unified citation layer across AI platforms, you're optimizing for one-third of the market and invisible to the rest.

The Citation Divergence: Platform by Platform

Here's what the research shows:

ChatGPT (GPT-5.5) — heavily weighted toward Wikipedia (34% of citations), major news sites (WSJ, NYT, Reuters at 18%), and academic databases (PubMed, arXiv at 12%). Content depth and citation count matter more than freshness.

Claude (Sonnet 4.5) — prioritizes niche authoritative sources (industry blogs, domain experts, technical documentation at 41%), virtually ignores Wikipedia (0.1%), and weights recent content heavily (72% of citations from content published within 90 days). Freshness is a primary ranking factor.

Perplexity (sonar-pro) — leans on real-time web scraping (53% of citations from content published within 14 days), Reddit and forum discussions (22%), and technical documentation (19%). Wikipedia citations: zero. Perplexity treats Wikipedia as outdated by design.

Why Wikipedia-First SEO Is a Losing Strategy

For years, the SEO playbook has been: “Get cited on Wikipedia. Earn featured snippets. Win knowledge panel placement.”

That still works — if your entire addressable market uses ChatGPT and Google AI Overviews.

But Claude and Perplexity users — which collectively represent 40% of generative search volume as of July 2026 — don't see Wikipedia citations. They see fresh content, niche authorities, and real-time discussions.

If you're optimizing for Wikipedia inclusion and ignoring platform-specific citation behaviors, you're invisible to 40% of your market.

The New GEO Playbook: Platform-Specific Strategies

A unified GEO strategy doesn't work anymore. You need three parallel optimization tracks:

For ChatGPT:

• Prioritize citation-dense content (10+ inline citations per 1000 words)
• Build relationships with Wikipedia editors to earn citations in relevant articles
• Target major news outlets for brand mentions and thought leadership
• Publish academically-structured content with clear methodology sections

For Claude:

• Maintain aggressive content refresh cadence (update every 30-60 days)
• Build deep niche authority in specific verticals (vs. broad topic coverage)
• Publish technical documentation, case studies, and practitioner-focused content
• Use visible “Last updated” timestamps and version history

For Perplexity:

• Publish real-time content (daily or weekly) to stay in the 14-day citation window
• Engage in Reddit, Hacker News, and industry forums where Perplexity scrapes heavily
• Maintain accessible technical documentation with clear, scrapable structure
• Optimize for recency signals (publication dates, event-driven content)

What This Means for Your Content Calendar

You can't optimize one piece of content for all three platforms. The citation behaviors are too divergent.

Instead, segment your content production:

Evergreen authority pieces (for ChatGPT) — in-depth guides, research summaries, academically-structured explainers. High citation density. Optimized for Wikipedia inclusion.

Niche practitioner content (for Claude) — technical deep-dives, case studies, implementation guides. Refreshed quarterly. Focused on specific verticals.

Real-time commentary (for Perplexity) — weekly updates, event-driven analysis, industry news takes. Published frequently. Promoted on Reddit and forums.

The Bottom Line

The generative search landscape isn't converging. It's diverging. ChatGPT, Claude, and Perplexity have fundamentally different citation behaviors — and those differences are getting wider, not narrower.

A single GEO strategy won't reach all three audiences. You need platform-specific optimization tracks, segmented content calendars, and measurement frameworks that track citation performance across all three platforms independently.

The businesses that treat GEO as a unified discipline will optimize for one platform and miss the other two. The ones that build platform-specific strategies will capture the full market.