Case Study: How We Improved Our GEO Score from 12 to 71

The Embarrassing Starting Point: 12/100

When we first ran our own tool against score.aeofy.ai, we got a 12 out of 100. A D grade. For a tool that literally scores websites on their AI search visibility, this was — to put it mildly — not a great look. Our website was a single-page Vite SPA with no server-side rendering, no structured data, no FAQ pages, and minimal content. AI search engines like ChatGPT, Perplexity, and Google AI Overviews had almost nothing to work with.

The GEO Scoring tool evaluates websites across 9 dimensions with specific weights: Citability (20 points), Schema Markup (25 points), Content Richness (20 points), Knowledge Graph Readiness (15 points), E-E-A-T Signals (10 points), Open Graph Protocol (5 points), and On-Page SEO (5 points). Each dimension measures a different aspect of how AI systems discover, understand, and cite your content. We were failing at almost all of them.

Round 1: Technical Foundation (12 → 42)

The first round focused entirely on technical infrastructure. We migrated from a Vite SPA to Nuxt 3 with server-side rendering. This single change was transformative — AI crawlers that previously saw an empty HTML shell could now read our actual content on first load. SSR isn't optional for GEO; it's the foundation everything else builds on.

What we added in Round 1:

• Nuxt 3 SSR migration — Every page now renders server-side with full HTML content
• Organization Schema — JSON-LD structured data identifying who we are
• WebSite Schema with SearchAction — Tells AI engines our site has a search/analysis function
• Open Graph + Twitter Card meta tags — Proper social sharing metadata on every page
• About page — Company information, mission, team context
• llms.txt — A machine-readable summary of our site for AI crawlers

Result: 12 → 42 points. Schema went from 0 to decent, OGP from 0 to full marks, and content started registering. But we were still weak on Citability and Content Richness.

Round 2: Content Structure (42 → 53)

Round 2 targeted Content Richness and Knowledge Graph dimensions. We created structured FAQ pages — not just a single FAQ dump, but organized topic pages covering GEO basics, schema markup guidance, and scoring methodology. Each FAQ page uses FAQPage Schema markup so AI engines can extract individual Q&A pairs.

What we added in Round 2:

• 3 FAQ topic pages — GEO Basics (8 questions), Schema Guide (6 questions), Scoring Methodology (6 questions)
• FAQ index page — Central hub linking to all FAQ topics
• HowTo guide — "How to Improve Your GEO Score" with 8 structured steps and HowTo Schema
• Footer navigation expansion — More internal links for crawlability
• Updated llms.txt — Added new pages to the AI-readable site summary

Result: 42 → 53 points. Content Richness improved significantly. The FAQ pages gave AI engines citable, structured answers. The HowTo guide added another Schema type, improving schema variety scoring.

Round 3: Citability & Authority (53 → 71)

The biggest jump came from focusing on Citability — the dimension that measures whether your content contains passages that AI systems would actually quote. Research from Princeton's GEO paper and our competitor analysis showed that AI systems prefer citing self-contained paragraphs between 100-200 words that include at least one concrete statistic or fact. We restructured our content specifically for this.

What we added in Round 3:

• 5 blog articles — Each 1500-2500 words with Article Schema, author attribution, and 100-200 word citable paragraphs
• Citability-optimized paragraphs — Every key paragraph is self-contained with specific data points
• Author attribution — Person Schema with sameAs links on all blog posts
• Brand authority signals — Expanded About page, external authority references
• Internal linking — Blog posts cross-reference each other and link to relevant FAQ/Guide pages

Result: 53 → 71 points. Citability jumped dramatically. The blog articles gave AI engines substantial, well-structured content to cite. E-E-A-T signals improved with author attribution.

What Moved the Needle Most

Looking back at the full journey from 12 to 71, three changes had the biggest individual impact on our GEO score. First, the Nuxt 3 SSR migration was worth approximately 20 points alone — without server-side rendering, AI crawlers simply cannot read JavaScript-rendered content. Second, structured FAQ pages with FAQPage Schema added roughly 10 points across Content Richness and Schema dimensions. Third, blog articles with citability-optimized paragraphs (100-200 words, each containing concrete statistics) drove another 15 points across Citability and E-E-A-T dimensions.

Key Takeaways for Your Website

Based on our experience optimizing from 12 to 71, here are the highest-impact actions any website can take to improve AI search visibility. Start with SSR or static site generation if you're using a JavaScript framework — this is non-negotiable. Add Organization and WebSite Schema markup on day one, then expand to FAQPage, HowTo, Article, and Product schemas as your content grows. Create FAQ pages with real, detailed answers to questions your audience actually asks. Write blog content with self-contained paragraphs of 100-200 words that include specific numbers and facts. Add author attribution to everything — AI systems evaluate content credibility through E-E-A-T signals.

What's Next: Targeting 80+

We're not done. Our target is an A grade (80+ points). The remaining gap is primarily in Schema depth (more nested schemas with detailed attributes), additional blog content (targeting 20+ articles), and Brand Authority signals. We're also exploring adding a Brand Authority scoring dimension inspired by competitor analysis. Follow our progress — you can scan our site anytime at score.aeofy.ai to see our current score.