Do AI Search Engines Actually Localize Results by Country?

We analyzed 56,223 citations across 4 countries and 6 AI search engines to see whether search results adapt to local markets - or if US-based sources dominate everywhere.

AI search engines promise personalized, location-aware results. But when we tracked how 6 major answer engines cite sources across different regions, the numbers told a different story.

The bottom line: 66.0% of all citations still come from global (primarily US-based) domains, regardless of what region users search from. Only 18.3% use proper country-code top-level domains (ccTLDs) that truly represent local markets.

This goes beyond search results. We wanted to see if AI engines are creating a more diverse information landscape or providing added value over traditional search for international users.

We analyzed 56,223 citations across 6 answer engines and 4 international markets. These citations were exported from the XFunnel database (July–August 2025). Citations were parsed and classified as either global (.com) or localized (ccTLDs like .fr, .de, .co.uk, or /fr/ subfolders). For non-English markets, queries were written in the local language (French, German, Dutch). The dataset focused on B2B SaaS and tech brands (many U.S.-headquartered), so .com domains may be over-represented. Our breakdown is simply meant to highlight how engines differ in surfacing local content when it exists.

These are the breakdowns we saw when AI search engines attempted localization:

Global Domains: 37,129 citations (66.0%)

• The default fallback for most queries
• Consistent across all regions and engines

ccTLD Domains: 10,271 citations (18.3%)

• Country-specific domains (.co.uk, .de, .au, etc.)
• Varies dramatically by region and engine

Subfolder Localization: 7,633 citations (13.6%)

• Site.com/fr/, site.com/uk/ style localization
• Mixed effectiveness across markets

Subdomain Localization: 1,190 citations (2.1%)

• fr.site.com, uk.site.com approaches
• Often overlooked by AI search engines

In classic Google search results, local country domains (ccTLDs) feature prominently – much more so than in the AI-generated citations we saw. For major European markets like France and Germany, the majority of top organic results typically come from local ccTLD sites.

For major European markets like France and Germany, the majority of top organic results typically come from local ccTLD sites.

For example, an analysis of Google's top 30 results (Page 1–3) found:

• Google.de → ~69.6% .de sites, only ~19.6% .com.
• Google.nl → ~70.7% .nl sites, ~20.8% .com.

In other words, traditional search heavily prioritizes local domains for users in these countries, far beyond what we saw for most answer engines' citations.

We took Google as a benchmark not because it's identical, but because it sets the user expectation for localization in search.

The variation between AI search engines in their localization efforts is massive - and reveals fundamentally different philosophies about serving global users.

The Localization Leaders

Perplexity (56.5% non-global citations) leads the pack in regional adaptation, consistently surfacing local sources and country-specific domains rather than defaulting to US sources. With 9,452 total citations, it shows the strongest commitment to finding region-relevant business information.

Microsoft Copilot (56.0% non-global citations) matches Perplexity's localization rate despite having fewer total citations (2,562). When users search from Germany for example, Copilot actively seeks out .de domains and German-language business sources.

The Middle Pack

Grok (36.2% non-global citations) demonstrates moderate regional awareness with 13,103 citations, particularly strong in emerging markets where it seeks out local business information rather than global defaults.

ChatGPT (29.7% non-global citations) shows lower localization effort across 7,325 citations, though still relies heavily on global sources for most queries. This isn't as surprising since it was trained on less regionalized data and as a result relies more heavily on global brands.

ChatGPT (+ Browsing) (28.6% non-global citations) handles the highest volume with 16,625 citations but shows inconsistent localization. Despite having browsing capability, it often defaults to global sources even when local alternatives exist.

The Global Defaulters

Gemini (5.3% non-global citations) demonstrates minimal localization effort across 1,730 citations, almost entirely relying on global domains regardless of user location. Honestly, I was shocked that Google didn't perform better here as it obviously has tons of relevant data to pull from.

Google AI Mode has only recently expanded to 180 countries and so it wasn't included in this analysis.

The data reveals stark differences in how AI search engines serve different global markets:

Localization By Country

Netherlands (54.5% non-global citations) leads global localization with 13,741 total citations. Dutch users see the highest rate of local domain citations, ccTLD usage, and regional business information across all engines.

Germany (44.6% non-global citations) ranks second with 13,725 citations, benefiting from strong local digital infrastructure and consistent AI engine attention to .de domains and German-language sources.

France (35.3% non-global citations) shows moderate localization across 13,839 citations, with decent ccTLD usage but room for improvement in regional source discovery.

UK (5.9% non-global citations) surprisingly ranks near the bottom despite its developed digital economy, with only 1,422 total citations showing minimal local domain preference - engines treat UK queries almost identically to US searches.

When we examined which types of sources reach the top citation positions, the localization gap becomes even more pronounced:

Global domains dominate top ranks: 66.5% of top citations (15,973 out of 24,013) come from global sources - actually slightly higher than their overall representation.

Local sources struggle for top positions: ccTLD domains drop from 18.3% overall to just 17.6% of top citations, while subfolder localization improves slightly to 14.9%.

Subdomain localization nearly invisible: The least common localization method drops to just 0.9% of top citations, making local subdomain sources almost impossible to find in AI results.

Next we checked if engines that localize overall, also localize the very top citation. Looking just at the top ranked citation:

• Perplexity 60.4% - Even stronger localization for the most prominent citation
• Copilot 50.2% - Less than their 56.0% average but still one of the most localized overall.
• Grok 30.2% - Drops slightly from its overall 36.2% rate
• ChatGPT (+ browsing) 24.9% - Lower than its overall 28.6% localization
• ChatGPT 22.0% - Drops from its overall 29.7% rate
• Gemini 1.2% - Even worse localization for top citations than overall

The engines that prioritize local sources do so most aggressively for their primary citation, while those that default to global sources become even more US-centric for their top recommendation.

The localization gap in answer engines creates real competitive implications:

Market Research Blind Spots: Companies researching new markets through AI may miss key local competitors and regulatory requirements - particularly problematic in regions like the UK where local sources represent less than 6% of citations.

Partner Discovery Bias: B2B buyers increasingly rely on AI for vendor research, but poor localization means local suppliers get systematically overlooked in favor of US-based alternatives.

Regional Competitive Advantage: Companies in the Netherlands and Germany benefit from AI search engines' relatively strong localization (54.5% and 44.6% respectively), while UK businesses face an uphill battle for AI visibility despite their developed market.

Customer Intelligence Failures: When 66% of all AI citations default to global sources, potential customers researching local solutions, compliance frameworks, and market-specific services miss critical regional information.

The 53-percentage-point difference between the best (Perplexity at 56.5%) and worst (Google Search at 3.4%) performing engines creates a fragmented global market where your choice of AI dramatically impacts the regional relevance of business information you receive.

For businesses: Monitor which answer engines your target customers use, as Perplexity and Copilot users see dramatically different local business representation than Gemini or Google Search users.

For market research: The engine you choose for competitive intelligence literally determines whether you discover local market players or remain blind to regional competition.

Want to learn all about optimizing your content for AI platforms, including a tailored content strategy for your company? Reach out to us at XFunnel and see how you can start tracking your brand's performance in GPT-5, Claude, Gemini, and lots of other AI platforms.

This analysis is based on real citation data from XFunnel's global localization monitoring system, tracking 56,223 citations across 6 answer engines and 4 international markets. The data represents actual business queries and source citations from AI search engines responding to location-specific searches.

About the Author

Jennifer Frame is the Director of Growth at XFunnel, an experienced SEO team leader with a background in advertising & marketing.

With 10+ years experience growing traffic & conversion rates for international sites, she leads strategic initiatives to help businesses optimize their visibility across AI search platforms.

Connect with her on LinkedIn.