We Analyzed 22,000+ AI Citations Across Latin America. Here's What Gets Cited.

The Experiment

Over two nights, we ran Tocho's citation pipeline against 22,000+ URLs across Latin America. For each URL, we asked Gemini: "If a user asked about this topic, would you cite this page?"

We weighted our sample toward the Americas:

• 40% Spanish — Mexico, Argentina, Colombia, Chile, Peru, Ecuador
• 30% Portuguese — Brazil
• 20% French — Quebec
• 10% English — Baseline

Here's what we found.

Overall: 37-40% of Pages Get Cited

Across all URLs in our Americas corpus, 37-40% were cited by Gemini. This is higher than the global average reported by other studies (~25-30%), likely because we prioritized authoritative regional content.

By Country: Colombia Leads, Chile Is a Ghost

| Country | Sites Cited | Observation | |---------|-----------|-------------| | Colombia | 127 | Highest citation rate. Colombian content is well-structured and covers niche topics. | | Brazil | 45 | Strong ecosystem. Lots of Portuguese content for AI to choose from. | | Mexico | 28 | Moderate. Mexico's top sites get cited, but mid-tier content is invisible. | | Quebec | 16 | Punches way above its weight. Low competition = high citation rate. | | Argentina | 14 | Underserved. Argentine content is strong but underrepresented in AI training. | | Chile | 1 | Almost zero. This is the biggest gap — and the biggest opportunity. |

Why Colombia Wins

Colombian content tends to be well-structured, bilingual-aware, and covers niche topics that other Spanish-speaking countries don't address. Sites like colombiafintech.co, colombiacheck.com, and icesi.edu.co are citation authorities in their fields.

Why Chile Is Missing

Only 1 Chilean site was cited across 1,300+ queries. This isn't because Chilean content is bad — it's because there's less of it in AI training data. For Chilean businesses, this is actually an opportunity: less competition means easier citation wins.

By Content Type: Reference Beats News

| Content Type | % of Citations | |-------------|---------------| | Niche authority (blogs, tools, SaaS) | 68% | | Education (.edu) | 9% | | Government (.gov) | 7% | | Tech/developer | 6% | | Finance/fintech | 4% | | E-commerce | 4% | | News | 2% |

The biggest surprise: news sites get cited only 2% of the time. AI prefers reference-quality content over breaking news.

What Does "Citable" Content Look Like?

From the sites that consistently got cited, we identified these patterns:

1. Clear, quotable definitions — "AI citation probability is defined as..."
2. Comparison tables — structured data AI can extract and present
3. FAQ sections — Q&A format maps directly to how AI generates responses
4. Specific numbers — "Revenue grew 23% in Q3 2025" beats vague claims
5. Named sources — "According to IBGE" gives AI a verifiable anchor
6. JSON-LD schema — pages with structured data get cited 28-36% more

What This Means for Your Content Strategy

1. If you're in Chile: You have almost zero competition for AI citations in Spanish. Start now and you own the space.
2. If you're in Colombia: Your content is already getting cited. Optimize to get cited MORE.
3. If you're in Quebec: French content has low competition. Bilingual optimization (FR + EN) is a superpower.
4. If you're writing news: Pivot to reference content. Write the guide, not the article.
5. If you're an agency: Your clients need to know that only 12% of AI citations match Google rankings. This is a new service you can sell.

Methodology

• Pipeline: Tocho citation runner with Gemini 2.5 Flash
• Sample: 22,000+ URLs from curated regional seed lists
• Weighting: ES 40%, PT 30%, FR 20%, EN 10%
• Score stratification: 40% high-scoring (70-100), 35% mid (40-69), 25% low (0-39)

Check Your Citation Probability

Want to know where your content stands? Check your AI citation probability for free →

Tocho predicts whether AI platforms will cite your content — per model, per language — trained on the same observation data used in this research.

Try it now →