Generative Engine Optimization

30 → 75 GEO Score in 30 Days

How I doubled an Indonesian charity's AI-search visibility for ChatGPT, Gemini, and Perplexity citations.

Global Ehsan Relief (Indonesia) 2026
30→75
GEO score (2.5×)
30
Day sprint
12
Phases shipped
$0
Paid tools needed

For Global Ehsan Relief, an Indonesian charity, a 30-day GEO sprint in 2026 took the site from a GEO score of 30 to 75 across 12 phases with no paid tools, so ChatGPT, Gemini and Perplexity could find and cite it.

The problem

The starting GEO score was 30/100 — the charity was invisible to AI citations. Schema markup was broken or missing on most pages, with no structured data for charity / donation entities. Meta descriptions were generic and keyword-sparse. No FAQ schema meant AI assistants couldn't extract Q&A pairs. There was no llms.txt to signal AI crawlers, and trust signals (UEN, IPC, charity registration number) were buried in the footer instead of surfaced for AI extraction.

Approach

Results

Traditional SEO ranks in Google blue links. GEO ranks in AI citations. Different signals, different playbook. Score before every phase to attribute lift. Ship one issue category per day, not parallel chaos. Document every fix as a reusable template for the next client.

Tech stack

Schema.org ValidatorPython (schema gen)Claude APITavily MCPGA4Search Consolellms.txt

Questions about this work

What was the starting point?

A GEO score of 30 out of 100. Schema markup was broken or missing on most pages, there was no FAQ schema and no llms.txt, meta descriptions were generic, and the charity's trust signals sat in the footer where AI systems do not look.

What moved the score?

Four phases. Foundation, plus 18 points: a full schema rebuild, canonical fixes, meta and H1 cleanup, trust signals in headers and Open Graph. Content, plus 12: FAQ schema, rebuilt service pages, social cards, llms.txt. Authority, plus 8: backlink audit, citations, a press kit, an About page rebuilt for authority. Monitoring, plus 7: AI citation tracking and a handover playbook.

What did it cost in tools?

$0 in paid tools. Custom Python generated the schema and Claude with structured prompts extracted the FAQ pairs.

Need this kind of work done?

Open to consulting and contract work in AI automation, ad operations, and GEO. Same approach, same metrics-driven rigor.

Work with me See other case studies