GEO Optimizer
A research-backed toolkit to make your website visible and citable by AI search engines.
The Problem Nobody Talks About
Traditional SEO focuses on ranking in Google's link-based results. But the search landscape has fundamentally shifted.
AI search engines don't show a list of links. They give a direct answer and cite their sources. When someone asks ChatGPT, Perplexity, Claude, or Gemini a question about your industry, your website is either cited in the answer or it is invisible. There is no page 2 -- there is only "cited" or "not cited."
This means that a website ranking #1 on Google can still be completely absent from AI-generated answers if it is not optimized for how AI systems discover, evaluate, and cite content.
Most businesses have no idea they are invisible to AI search. Their SEO metrics look fine, their Google rankings are solid, but AI assistants never mention them. This is the GEO gap.
What is GEO?
Generative Engine Optimization (GEO) is the practice of optimizing web content so that AI-powered search engines cite it in their generated responses. The term and methodology originate from a Princeton University research paper presented at KDD 2024, one of the premier conferences in data mining and knowledge discovery.
The study, "GEO: Generative Engine Optimization", tested nine distinct content optimization methods across 10,000 real queries on Perplexity.ai and measured which techniques actually increase a website's visibility in AI-generated answers.
Key Research Findings
The results are striking:
| Method | Visibility Improvement |
|---|---|
| Cite Sources | +30 to +115% |
| Statistics | +40% average |
| Quotation Addition | +30 to +40% |
| Fluency Optimization | +15 to +30% |
| Easy-to-Understand | +8 to +15% |
| Authoritative Tone | +6 to +12% |
| Technical Terms | +5 to +10% |
| Unique Words | +5 to +8% |
| Keyword Stuffing | ~0% (ineffective) |
The top method -- adding inline source citations -- can more than double a page's visibility in AI search results. Meanwhile, keyword stuffing, a holdover from traditional SEO, has zero or even negative impact.
What's Inside the Toolkit
GEO Optimizer provides three automation scripts and a collection of AI context files that implement the Princeton methodology:
Core Scripts
| Script | Purpose |
|---|---|
geo_audit.py | Scores your website 0--100 across five GEO dimensions (robots.txt, llms.txt, JSON-LD schema, meta tags, content quality) |
generate_llms_txt.py | Auto-generates an /llms.txt file from your sitemap, making your site structure machine-readable for AI crawlers |
schema_injector.py | Generates, validates, and injects JSON-LD structured data (WebSite, FAQPage, Article, Organization, and more) |
AI Context Files
Pre-built configuration files for six AI platforms, so you can use GEO Optimizer as a skill inside your preferred AI assistant:
- Claude Projects -- full context (~11,700 chars)
- ChatGPT Custom GPT -- optimized for 8k char limit
- ChatGPT Custom Instructions -- compressed for 1.5k char limit
- Cursor -- IDE rules file
- Windsurf -- IDE rules file
- Kiro -- steering file
Quality Metrics
- 89 tests (67 unit + 13 integration + 9 validation)
- 87% business logic coverage, 70% total
- Quality score: 9.25/10 (enterprise-grade)
- Python 3.8+ compatibility
How It Works
The typical workflow follows four steps:
1. Audit --> Run geo_audit.py to get your current score
2. Fix --> Address infrastructure gaps (robots.txt, llms.txt, schema)
3. Optimize --> Refine content using the 9 Princeton methods
4. Retest --> Run the audit again to measure improvement
If you want to jump straight in, head to the Installation guide. You can have your first audit running in under two minutes.
Who Is This For
- Web developers building sites that need AI visibility
- SEO professionals expanding into generative engine optimization
- Content teams looking for data-driven content improvement signals
- Agencies offering AI search optimization as a service
License
GEO Optimizer is open source under the MIT License.
Next: Installation -- get GEO Optimizer running on your machine.